id
large_string | submitter
large_string | authors
large_string | title
large_string | comments
large_string | journal-ref
large_string | doi
large_string | report-no
large_string | categories
large_string | license
large_string | abstract
large_string | versions
large list | update_date
timestamp[ms] | authors_parsed
large list | prompt
large_string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1501.00436 | Sebasti\'an Basterrech | Sebasti\'an Basterrech and Enrique Alba and V\'aclav Sn\'a\v{s}el | An Experimental Analysis of the Echo State Network Initialization Using
the Particle Swarm Optimization | IEEE Sixth World Congress on Nature and Biologically Inspired
Computing (NaBIC2014), Porto, Portugal, July 30-August 1, 2014 | null | null | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This article introduces a robust hybrid method for solving supervised
learning tasks, which uses the Echo State Network (ESN) model and the Particle
Swarm Optimization (PSO) algorithm. An ESN is a Recurrent Neural Network with
the hidden-hidden weights fixed in the learning process. The recurrent part of
the network stores the input information in internal states of the network.
Another structure forms a free-memory method used as supervised learning tool.
The setting procedure for initializing the recurrent structure of the ESN model
can impact on the model performance. On the other hand, the PSO has been shown
to be a successful technique for finding optimal points in complex spaces.
Here, we present an approach to use the PSO for finding some initial
hidden-hidden weights of the ESN model. We present empirical results that
compare the canonical ESN model with this hybrid method on a wide range of
benchmark problems.
| [
{
"version": "v1",
"created": "Fri, 2 Jan 2015 16:49:29 GMT"
}
] | 2015-01-05T00:00:00 | [
[
"Basterrech",
"Sebastián",
""
],
[
"Alba",
"Enrique",
""
],
[
"Snášel",
"Václav",
""
]
] | TITLE: An Experimental Analysis of the Echo State Network Initialization Using
the Particle Swarm Optimization
ABSTRACT: This article introduces a robust hybrid method for solving supervised
learning tasks, which uses the Echo State Network (ESN) model and the Particle
Swarm Optimization (PSO) algorithm. An ESN is a Recurrent Neural Network with
the hidden-hidden weights fixed in the learning process. The recurrent part of
the network stores the input information in internal states of the network.
Another structure forms a free-memory method used as supervised learning tool.
The setting procedure for initializing the recurrent structure of the ESN model
can impact on the model performance. On the other hand, the PSO has been shown
to be a successful technique for finding optimal points in complex spaces.
Here, we present an approach to use the PSO for finding some initial
hidden-hidden weights of the ESN model. We present empirical results that
compare the canonical ESN model with this hybrid method on a wide range of
benchmark problems.
|
1407.6251 | Philip Lenz | Philip Lenz, Andreas Geiger, Raquel Urtasun | FollowMe: Efficient Online Min-Cost Flow Tracking with Bounded Memory
and Computation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | One of the most popular approaches to multi-target tracking is
tracking-by-detection. Current min-cost flow algorithms which solve the data
association problem optimally have three main drawbacks: they are
computationally expensive, they assume that the whole video is given as a
batch, and they scale badly in memory and computation with the length of the
video sequence. In this paper, we address each of these issues, resulting in a
computationally and memory-bounded solution. First, we introduce a dynamic
version of the successive shortest-path algorithm which solves the data
association problem optimally while reusing computation, resulting in
significantly faster inference than standard solvers. Second, we address the
optimal solution to the data association problem when dealing with an incoming
stream of data (i.e., online setting). Finally, we present our main
contribution which is an approximate online solution with bounded memory and
computation which is capable of handling videos of arbitrarily length while
performing tracking in real time. We demonstrate the effectiveness of our
algorithms on the KITTI and PETS2009 benchmarks and show state-of-the-art
performance, while being significantly faster than existing solvers.
| [
{
"version": "v1",
"created": "Wed, 23 Jul 2014 15:07:12 GMT"
},
{
"version": "v2",
"created": "Thu, 25 Dec 2014 14:15:01 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Lenz",
"Philip",
""
],
[
"Geiger",
"Andreas",
""
],
[
"Urtasun",
"Raquel",
""
]
] | TITLE: FollowMe: Efficient Online Min-Cost Flow Tracking with Bounded Memory
and Computation
ABSTRACT: One of the most popular approaches to multi-target tracking is
tracking-by-detection. Current min-cost flow algorithms which solve the data
association problem optimally have three main drawbacks: they are
computationally expensive, they assume that the whole video is given as a
batch, and they scale badly in memory and computation with the length of the
video sequence. In this paper, we address each of these issues, resulting in a
computationally and memory-bounded solution. First, we introduce a dynamic
version of the successive shortest-path algorithm which solves the data
association problem optimally while reusing computation, resulting in
significantly faster inference than standard solvers. Second, we address the
optimal solution to the data association problem when dealing with an incoming
stream of data (i.e., online setting). Finally, we present our main
contribution which is an approximate online solution with bounded memory and
computation which is capable of handling videos of arbitrarily length while
performing tracking in real time. We demonstrate the effectiveness of our
algorithms on the KITTI and PETS2009 benchmarks and show state-of-the-art
performance, while being significantly faster than existing solvers.
|
1412.7242 | Chengyao Shen | Chengyao Shen, Xun Huang and Qi Zhao | Learning of Proto-object Representations via Fixations on Low Resolution | This paper has been withdrawn by the author due to incompletion of
the submission | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While previous researches in eye fixation prediction typically rely on
integrating low-level features (e.g. color, edge) to form a saliency map,
recently it has been found that the structural organization of these features
into a proto-object representation can play a more significant role. In this
work, we present a computational framework based on deep network to demonstrate
that proto-object representations can be learned from low-resolution image
patches from fixation regions. We advocate the use of low-resolution inputs in
this work due to the following reasons: (1) Proto-objects are computed in
parallel over an entire visual field (2) People can perceive or recognize
objects well even it is in low resolution. (3) Fixations from lower resolution
images can predict fixations on higher resolution images. In the proposed
computational model, we extract multi-scale image patches on fixation regions
from eye fixation datasets, resize them to low resolution and feed them into a
hierarchical. With layer-wise unsupervised feature learning, we find that many
proto-objects like features responsive to different shapes of object blobs are
learned out. Visualizations also show that these features are selective to
potential objects in the scene and the responses of these features work well in
predicting eye fixations on the images when combined with learned weights.
| [
{
"version": "v1",
"created": "Tue, 23 Dec 2014 03:14:21 GMT"
},
{
"version": "v2",
"created": "Sat, 27 Dec 2014 08:29:00 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Shen",
"Chengyao",
""
],
[
"Huang",
"Xun",
""
],
[
"Zhao",
"Qi",
""
]
] | TITLE: Learning of Proto-object Representations via Fixations on Low Resolution
ABSTRACT: While previous researches in eye fixation prediction typically rely on
integrating low-level features (e.g. color, edge) to form a saliency map,
recently it has been found that the structural organization of these features
into a proto-object representation can play a more significant role. In this
work, we present a computational framework based on deep network to demonstrate
that proto-object representations can be learned from low-resolution image
patches from fixation regions. We advocate the use of low-resolution inputs in
this work due to the following reasons: (1) Proto-objects are computed in
parallel over an entire visual field (2) People can perceive or recognize
objects well even it is in low resolution. (3) Fixations from lower resolution
images can predict fixations on higher resolution images. In the proposed
computational model, we extract multi-scale image patches on fixation regions
from eye fixation datasets, resize them to low resolution and feed them into a
hierarchical. With layer-wise unsupervised feature learning, we find that many
proto-objects like features responsive to different shapes of object blobs are
learned out. Visualizations also show that these features are selective to
potential objects in the scene and the responses of these features work well in
predicting eye fixations on the images when combined with learned weights.
|
1412.7782 | Roshan Ragel | MAC Jiffriya, MAC Akmal Jahan, and Roshan G. Ragel | Plagiarism Detection on Electronic Text based Assignments using Vector
Space Model (ICIAfS14) | appears in The 7th International Conference on Information and
Automation for Sustainability (ICIAfS) 2014 | null | null | null | cs.IR cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Plagiarism is known as illegal use of others' part of work or whole work as
one's own in any field such as art, poetry, literature, cinema, research and
other creative forms of study. Plagiarism is one of the important issues in
academic and research fields and giving more concern in academic systems. The
situation is even worse with the availability of ample resources on the web.
This paper focuses on an effective plagiarism detection tool on identifying
suitable intra-corpal plagiarism detection for text based assignments by
comparing unigram, bigram, trigram of vector space model with cosine similarity
measure. Manually evaluated, labelled dataset was tested using unigram, bigram
and trigram vector. Even though trigram vector consumes comparatively more
time, it shows better results with the labelled data. In addition, the selected
trigram vector space model with cosine similarity measure is compared with
tri-gram sequence matching technique with Jaccard measure. In the results,
cosine similarity score shows slightly higher values than the other. Because,
it focuses on giving more weight for terms that do not frequently exist in the
dataset and cosine similarity measure using trigram technique is more
preferable than the other. Therefore, we present our new tool and it could be
used as an effective tool to evaluate text based electronic assignments and
minimize the plagiarism among students.
| [
{
"version": "v1",
"created": "Thu, 25 Dec 2014 03:54:01 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Jiffriya",
"MAC",
""
],
[
"Jahan",
"MAC Akmal",
""
],
[
"Ragel",
"Roshan G.",
""
]
] | TITLE: Plagiarism Detection on Electronic Text based Assignments using Vector
Space Model (ICIAfS14)
ABSTRACT: Plagiarism is known as illegal use of others' part of work or whole work as
one's own in any field such as art, poetry, literature, cinema, research and
other creative forms of study. Plagiarism is one of the important issues in
academic and research fields and giving more concern in academic systems. The
situation is even worse with the availability of ample resources on the web.
This paper focuses on an effective plagiarism detection tool on identifying
suitable intra-corpal plagiarism detection for text based assignments by
comparing unigram, bigram, trigram of vector space model with cosine similarity
measure. Manually evaluated, labelled dataset was tested using unigram, bigram
and trigram vector. Even though trigram vector consumes comparatively more
time, it shows better results with the labelled data. In addition, the selected
trigram vector space model with cosine similarity measure is compared with
tri-gram sequence matching technique with Jaccard measure. In the results,
cosine similarity score shows slightly higher values than the other. Because,
it focuses on giving more weight for terms that do not frequently exist in the
dataset and cosine similarity measure using trigram technique is more
preferable than the other. Therefore, we present our new tool and it could be
used as an effective tool to evaluate text based electronic assignments and
minimize the plagiarism among students.
|
1412.7789 | Roshan Ragel | Vajira Thambawita, Roshan Ragel and Dhammika Elkaduwe | To Use or Not to Use: Graphics Processing Units for Pattern Matching
Algorithms | appears in The 7th International Conference on Information and
Automation for Sustainability (ICIAfS) 2014 | null | null | null | cs.DC cs.PF | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | String matching is an important part in today's computer applications and
Aho-Corasick algorithm is one of the main string matching algorithms used to
accomplish this. This paper discusses that when can the GPUs be used for string
matching applications using the Aho-Corasick algorithm as a benchmark. We have
to identify the best unit to run our string matching algorithm according to the
performance of our devices and the applications. Sometimes CPU gives better
performance than GPU and sometimes GPU gives better performance than CPU.
Therefore, identifying this critical point is significant task for researchers
who are using GPUs to improve the performance of their string matching
applications based on string matching algorithms.
| [
{
"version": "v1",
"created": "Thu, 25 Dec 2014 05:27:49 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Thambawita",
"Vajira",
""
],
[
"Ragel",
"Roshan",
""
],
[
"Elkaduwe",
"Dhammika",
""
]
] | TITLE: To Use or Not to Use: Graphics Processing Units for Pattern Matching
Algorithms
ABSTRACT: String matching is an important part in today's computer applications and
Aho-Corasick algorithm is one of the main string matching algorithms used to
accomplish this. This paper discusses that when can the GPUs be used for string
matching applications using the Aho-Corasick algorithm as a benchmark. We have
to identify the best unit to run our string matching algorithm according to the
performance of our devices and the applications. Sometimes CPU gives better
performance than GPU and sometimes GPU gives better performance than CPU.
Therefore, identifying this critical point is significant task for researchers
who are using GPUs to improve the performance of their string matching
applications based on string matching algorithms.
|
1412.7851 | Odemir Bruno PhD | Jo\~ao Batista Florindo and Odemir Martinez Bruno | Fractal descriptors based on the probability dimension: a texture
analysis and classification approach | 7 pages, 6 figures. arXiv admin note: text overlap with
arXiv:1205.2821 | Pattern Recognition Letters, Volume 42, Pages 107-114, 2014 | 10.1016/j.patrec.2014.01.009 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we propose a novel technique for obtaining descriptors of
gray-level texture images. The descriptors are provided by applying a
multiscale transform to the fractal dimension of the image estimated through
the probability (Voss) method. The effectiveness of the descriptors is verified
in a classification task using benchmark over texture datasets. The results
obtained demonstrate the efficiency of the proposed method as a tool for the
description and discrimination of texture images.
| [
{
"version": "v1",
"created": "Thu, 25 Dec 2014 18:50:31 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Florindo",
"João Batista",
""
],
[
"Bruno",
"Odemir Martinez",
""
]
] | TITLE: Fractal descriptors based on the probability dimension: a texture
analysis and classification approach
ABSTRACT: In this work, we propose a novel technique for obtaining descriptors of
gray-level texture images. The descriptors are provided by applying a
multiscale transform to the fractal dimension of the image estimated through
the probability (Voss) method. The effectiveness of the descriptors is verified
in a classification task using benchmark over texture datasets. The results
obtained demonstrate the efficiency of the proposed method as a tool for the
description and discrimination of texture images.
|
1412.7880 | Odemir Bruno PhD | Marcos W. S. Oliveira and Dalcimar Casanova and Jo\~ao B. Florindo and
Odemir Martinez Bruno | Enhancing fractal descriptors on images by combining boundary and
interior of Minkowski dilation | 6 pages 3 figures | Physica A, Volume 416, Pages 41-48, 2014 | 10.1016/j.physa.2014.07.074 | null | physics.data-an cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work proposes to obtain novel fractal descriptors from gray-level
texture images by combining information from interior and boundary measures of
the Minkowski dilation applied to the texture surface. At first, the image is
converted into a surface where the height of each point is the gray intensity
of the respective pixel in that position in the image. Thus, this surface is
morphologically dilated by spheres. The radius of such spheres is ranged within
an interval and the volume and the external area of the dilated structure are
computed for each radius. The final descriptors are given by such measures
concatenated and subject to a canonical transform to reduce the dimensionality.
The proposal is an enhancement to the classical Bouligand-Minkowski fractal
descriptors, where only the volume (interior) information is considered. As
different structures may have the same volume, but not the same area, the
proposal yields to more rich descriptors as confirmed by results on the
classification of benchmark databases.
| [
{
"version": "v1",
"created": "Fri, 26 Dec 2014 02:13:04 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Oliveira",
"Marcos W. S.",
""
],
[
"Casanova",
"Dalcimar",
""
],
[
"Florindo",
"João B.",
""
],
[
"Bruno",
"Odemir Martinez",
""
]
] | TITLE: Enhancing fractal descriptors on images by combining boundary and
interior of Minkowski dilation
ABSTRACT: This work proposes to obtain novel fractal descriptors from gray-level
texture images by combining information from interior and boundary measures of
the Minkowski dilation applied to the texture surface. At first, the image is
converted into a surface where the height of each point is the gray intensity
of the respective pixel in that position in the image. Thus, this surface is
morphologically dilated by spheres. The radius of such spheres is ranged within
an interval and the volume and the external area of the dilated structure are
computed for each radius. The final descriptors are given by such measures
concatenated and subject to a canonical transform to reduce the dimensionality.
The proposal is an enhancement to the classical Bouligand-Minkowski fractal
descriptors, where only the volume (interior) information is considered. As
different structures may have the same volume, but not the same area, the
proposal yields to more rich descriptors as confirmed by results on the
classification of benchmark databases.
|
1412.7963 | Odemir Bruno PhD | Jo\~ao B. Florindo, Odemir M. Bruno | Texture analysis by multi-resolution fractal descriptors | 8 pages, 6 figures | Expert Systems with Applications, Volume 40, Issue 10, Pages
4022-4028, 2013 | 10.1016/j.eswa.2013.01.007 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work proposes a texture descriptor based on fractal theory. The method
is based on the Bouligand-Minkowski descriptors. We decompose the original
image recursively into 4 equal parts. In each recursion step, we estimate the
average and the deviation of the Bouligand-Minkowski descriptors computed over
each part. Thus, we extract entropy features from both average and deviation.
The proposed descriptors are provided by the concatenation of such measures.
The method is tested in a classification experiment under well known datasets,
that is, Brodatz and Vistex. The results demonstrate that the proposed
technique achieves better results than classical and state-of-the-art texture
descriptors, such as Gabor-wavelets and co-occurrence matrix.
| [
{
"version": "v1",
"created": "Fri, 26 Dec 2014 17:45:41 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Florindo",
"João B.",
""
],
[
"Bruno",
"Odemir M.",
""
]
] | TITLE: Texture analysis by multi-resolution fractal descriptors
ABSTRACT: This work proposes a texture descriptor based on fractal theory. The method
is based on the Bouligand-Minkowski descriptors. We decompose the original
image recursively into 4 equal parts. In each recursion step, we estimate the
average and the deviation of the Bouligand-Minkowski descriptors computed over
each part. Thus, we extract entropy features from both average and deviation.
The proposed descriptors are provided by the concatenation of such measures.
The method is tested in a classification experiment under well known datasets,
that is, Brodatz and Vistex. The results demonstrate that the proposed
technique achieves better results than classical and state-of-the-art texture
descriptors, such as Gabor-wavelets and co-occurrence matrix.
|
1412.7990 | Ernesto Diaz-Aviles | Ernesto Diaz-Aviles, Hoang Thanh Lam, Fabio Pinelli, Stefano Braghin,
Yiannis Gkoufas, Michele Berlingerio, and Francesco Calabrese | Predicting User Engagement in Twitter with Collaborative Ranking | RecSysChallenge'14 at RecSys 2014, October 10, 2014, Foster City, CA,
USA | In Proceedings of the 2014 Recommender Systems Challenge
(RecSysChallenge'14). ACM, New York, NY, USA, , Pages 41 , 6 pages | 10.1145/2668067.2668072 | null | cs.IR cs.CY cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Collaborative Filtering (CF) is a core component of popular web-based
services such as Amazon, YouTube, Netflix, and Twitter. Most applications use
CF to recommend a small set of items to the user. For instance, YouTube
presents to a user a list of top-n videos she would likely watch next based on
her rating and viewing history. Current methods of CF evaluation have been
focused on assessing the quality of a predicted rating or the ranking
performance for top-n recommended items. However, restricting the recommender
system evaluation to these two aspects is rather limiting and neglects other
dimensions that could better characterize a well-perceived recommendation. In
this paper, instead of optimizing rating or top-n recommendation, we focus on
the task of predicting which items generate the highest user engagement. In
particular, we use Twitter as our testbed and cast the problem as a
Collaborative Ranking task where the rich features extracted from the metadata
of the tweets help to complement the transaction information limited to user
ids, item ids, ratings and timestamps. We learn a scoring function that
directly optimizes the user engagement in terms of nDCG@10 on the predicted
ranking. Experiments conducted on an extended version of the MovieTweetings
dataset, released as part of the RecSys Challenge 2014, show the effectiveness
of our approach.
| [
{
"version": "v1",
"created": "Fri, 26 Dec 2014 21:00:14 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Diaz-Aviles",
"Ernesto",
""
],
[
"Lam",
"Hoang Thanh",
""
],
[
"Pinelli",
"Fabio",
""
],
[
"Braghin",
"Stefano",
""
],
[
"Gkoufas",
"Yiannis",
""
],
[
"Berlingerio",
"Michele",
""
],
[
"Calabrese",
"Francesco",
""
]
] | TITLE: Predicting User Engagement in Twitter with Collaborative Ranking
ABSTRACT: Collaborative Filtering (CF) is a core component of popular web-based
services such as Amazon, YouTube, Netflix, and Twitter. Most applications use
CF to recommend a small set of items to the user. For instance, YouTube
presents to a user a list of top-n videos she would likely watch next based on
her rating and viewing history. Current methods of CF evaluation have been
focused on assessing the quality of a predicted rating or the ranking
performance for top-n recommended items. However, restricting the recommender
system evaluation to these two aspects is rather limiting and neglects other
dimensions that could better characterize a well-perceived recommendation. In
this paper, instead of optimizing rating or top-n recommendation, we focus on
the task of predicting which items generate the highest user engagement. In
particular, we use Twitter as our testbed and cast the problem as a
Collaborative Ranking task where the rich features extracted from the metadata
of the tweets help to complement the transaction information limited to user
ids, item ids, ratings and timestamps. We learn a scoring function that
directly optimizes the user engagement in terms of nDCG@10 on the predicted
ranking. Experiments conducted on an extended version of the MovieTweetings
dataset, released as part of the RecSys Challenge 2014, show the effectiveness
of our approach.
|
1412.8070 | Artiom Kovnatsky Artiom Kovnatsky | Artiom Kovnatsky, Michael M. Bronstein, Xavier Bresson, Pierre
Vandergheynst | Functional correspondence by matrix completion | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we consider the problem of finding dense intrinsic
correspondence between manifolds using the recently introduced functional
framework. We pose the functional correspondence problem as matrix completion
with manifold geometric structure and inducing functional localization with the
$L_1$ norm. We discuss efficient numerical procedures for the solution of our
problem. Our method compares favorably to the accuracy of state-of-the-art
correspondence algorithms on non-rigid shape matching benchmarks, and is
especially advantageous in settings when only scarce data is available.
| [
{
"version": "v1",
"created": "Sat, 27 Dec 2014 17:53:06 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Kovnatsky",
"Artiom",
""
],
[
"Bronstein",
"Michael M.",
""
],
[
"Bresson",
"Xavier",
""
],
[
"Vandergheynst",
"Pierre",
""
]
] | TITLE: Functional correspondence by matrix completion
ABSTRACT: In this paper, we consider the problem of finding dense intrinsic
correspondence between manifolds using the recently introduced functional
framework. We pose the functional correspondence problem as matrix completion
with manifold geometric structure and inducing functional localization with the
$L_1$ norm. We discuss efficient numerical procedures for the solution of our
problem. Our method compares favorably to the accuracy of state-of-the-art
correspondence algorithms on non-rigid shape matching benchmarks, and is
especially advantageous in settings when only scarce data is available.
|
1412.8099 | Rathipriya R | R. Rathipriya, K. Thangavel | Extraction of Web Usage Profiles using Simulated Annealing Based
Biclustering Approach | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, the Simulated Annealing (SA) based biclustering approach is
proposed in which SA is used as an optimization tool for biclustering of web
usage data to identify the optimal user profile from the given web usage data.
Extracted biclusters are consists of correlated users whose usage behaviors are
similar across the subset of web pages of a web site where as these users are
uncorrelated for remaining pages of a web site. These results are very useful
in web personalization so that it communicates better with its users and for
making customized prediction. Also useful for providing customized web service
too. Experiment was conducted on the real web usage dataset called CTI dataset.
Results show that proposed SA based biclustering approach can extract highly
correlated user groups from the preprocessed web usage data.
| [
{
"version": "v1",
"created": "Mon, 1 Dec 2014 10:06:25 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Rathipriya",
"R.",
""
],
[
"Thangavel",
"K.",
""
]
] | TITLE: Extraction of Web Usage Profiles using Simulated Annealing Based
Biclustering Approach
ABSTRACT: In this paper, the Simulated Annealing (SA) based biclustering approach is
proposed in which SA is used as an optimization tool for biclustering of web
usage data to identify the optimal user profile from the given web usage data.
Extracted biclusters are consists of correlated users whose usage behaviors are
similar across the subset of web pages of a web site where as these users are
uncorrelated for remaining pages of a web site. These results are very useful
in web personalization so that it communicates better with its users and for
making customized prediction. Also useful for providing customized web service
too. Experiment was conducted on the real web usage dataset called CTI dataset.
Results show that proposed SA based biclustering approach can extract highly
correlated user groups from the preprocessed web usage data.
|
1412.8104 | Natarajan Meghanathan | Natarajan Meghanathan | Performance Comparison of Minimum Hop vs. Minimum Edge Based Multicast
Routing under Different Mobility Models for Mobile Ad hoc Networks | 14 pages, 17 figures | International Journal of Wireless and Mobile Networks, 3(3):1-14,
June 2011 | null | null | cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The high-level contribution of this paper is to establish benchmarks for the
minimum hop count per source-receiver path and the minimum number of edges per
tree for multicast routing in mobile ad hoc networks (MANETs) under different
mobility models. In this pursuit, we explore the tradeoffs between these two
routing strategies with respect to hop count, number of edges and lifetime per
multicast tree with respect to the Random Waypoint, City Section and Manhattan
mobility models. We employ the Breadth First Search algorithm and the Minimum
Steiner Tree heuristic for determining a sequence of minimum hop and minimum
edge trees respectively. While both the minimum hop and minimum edge trees
exist for a relatively longer time under the Manhattan mobility model; the
number of edges per tree and the hop count per source-receiver path are
relatively low under the Random Waypoint model. For all the three mobility
models, the minimum edge trees have a longer lifetime compared to the minimum
hop trees and the difference in lifetime increases with increase in network
density and/or the multicast group size. Multicast trees determined under the
City Section model incur fewer edges and lower hop count compared to the
Manhattan mobility model.
| [
{
"version": "v1",
"created": "Sun, 7 Dec 2014 20:48:15 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Meghanathan",
"Natarajan",
""
]
] | TITLE: Performance Comparison of Minimum Hop vs. Minimum Edge Based Multicast
Routing under Different Mobility Models for Mobile Ad hoc Networks
ABSTRACT: The high-level contribution of this paper is to establish benchmarks for the
minimum hop count per source-receiver path and the minimum number of edges per
tree for multicast routing in mobile ad hoc networks (MANETs) under different
mobility models. In this pursuit, we explore the tradeoffs between these two
routing strategies with respect to hop count, number of edges and lifetime per
multicast tree with respect to the Random Waypoint, City Section and Manhattan
mobility models. We employ the Breadth First Search algorithm and the Minimum
Steiner Tree heuristic for determining a sequence of minimum hop and minimum
edge trees respectively. While both the minimum hop and minimum edge trees
exist for a relatively longer time under the Manhattan mobility model; the
number of edges per tree and the hop count per source-receiver path are
relatively low under the Random Waypoint model. For all the three mobility
models, the minimum edge trees have a longer lifetime compared to the minimum
hop trees and the difference in lifetime increases with increase in network
density and/or the multicast group size. Multicast trees determined under the
City Section model incur fewer edges and lower hop count compared to the
Manhattan mobility model.
|
1412.8118 | Lanbo Zhang | Lanbo Zhang and Yi Zhang | Hierarchical Bayesian Models with Factorization for Content-Based
Recommendation | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most existing content-based filtering approaches learn user profiles
independently without capturing the similarity among users. Bayesian
hierarchical models \cite{Zhang:Efficient} learn user profiles jointly and have
the advantage of being able to borrow discriminative information from other
users through a Bayesian prior. However, the standard Bayesian hierarchical
models assume all user profiles are generated from the same prior. Considering
the diversity of user interests, this assumption could be improved by
introducing more flexibility. Besides, most existing content-based filtering
approaches implicitly assume that each user profile corresponds to exactly one
user interest and fail to capture a user's multiple interests (information
needs).
In this paper, we present a flexible Bayesian hierarchical modeling approach
to model both commonality and diversity among users as well as individual
users' multiple interests. We propose two models each with different
assumptions, and the proposed models are called Discriminative Factored Prior
Models (DFPM). In our models, each user profile is modeled as a discriminative
classifier with a factored model as its prior, and different factors contribute
in different levels to each user profile. Compared with existing content-based
filtering models, DFPM are interesting because they can 1) borrow
discriminative criteria of other users while learning a particular user profile
through the factored prior; 2) trade off well between diversity and commonality
among users; and 3) handle the challenging classification situation where each
class contains multiple concepts. The experimental results on a dataset
collected from real users on digg.com show that our models significantly
outperform the baseline models of L-2 regularized logistic regression and
traditional Bayesian hierarchical model with logistic regression.
| [
{
"version": "v1",
"created": "Sun, 28 Dec 2014 06:07:48 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Zhang",
"Lanbo",
""
],
[
"Zhang",
"Yi",
""
]
] | TITLE: Hierarchical Bayesian Models with Factorization for Content-Based
Recommendation
ABSTRACT: Most existing content-based filtering approaches learn user profiles
independently without capturing the similarity among users. Bayesian
hierarchical models \cite{Zhang:Efficient} learn user profiles jointly and have
the advantage of being able to borrow discriminative information from other
users through a Bayesian prior. However, the standard Bayesian hierarchical
models assume all user profiles are generated from the same prior. Considering
the diversity of user interests, this assumption could be improved by
introducing more flexibility. Besides, most existing content-based filtering
approaches implicitly assume that each user profile corresponds to exactly one
user interest and fail to capture a user's multiple interests (information
needs).
In this paper, we present a flexible Bayesian hierarchical modeling approach
to model both commonality and diversity among users as well as individual
users' multiple interests. We propose two models each with different
assumptions, and the proposed models are called Discriminative Factored Prior
Models (DFPM). In our models, each user profile is modeled as a discriminative
classifier with a factored model as its prior, and different factors contribute
in different levels to each user profile. Compared with existing content-based
filtering models, DFPM are interesting because they can 1) borrow
discriminative criteria of other users while learning a particular user profile
through the factored prior; 2) trade off well between diversity and commonality
among users; and 3) handle the challenging classification situation where each
class contains multiple concepts. The experimental results on a dataset
collected from real users on digg.com show that our models significantly
outperform the baseline models of L-2 regularized logistic regression and
traditional Bayesian hierarchical model with logistic regression.
|
1412.8120 | Amlan Kusum | Amlan Kusum, Iulian Neamtiu and Rajiv Gupta | Adapting Graph Application Performance via Alternate Data Structure
Representation | Part of ADAPT Workshop proceedings, 2015 (arXiv:1412.2347) | null | null | ADAPT/2015/03 | cs.PL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph processing is used extensively in areas from social networking mining
to web indexing. We demonstrate that the performance and dependability of such
applications critically hinges on the graph data structure used, because a
fixed, compile-time choice of data structure can lead to poor performance or
applications unable to complete. To address this problem, we introduce an
approach that helps programmers transform regular, off-the-shelf graph
applications into adaptive, more dependable applications where adaptations are
performed via runtime selection from alternate data structure representations.
Using our approach, applications dynamically adapt to the input graph's
characteristics and changes in available memory so they continue to run when
faced with adverse conditions such as low memory. Experiments with graph
algorithms on real-world (e.g., Wikipedia metadata, Gnutella topology) and
synthetic graph datasets show that our adaptive applications run to completion
with lower execution time and/or memory utilization in comparison to their
non-adaptive versions.
| [
{
"version": "v1",
"created": "Sun, 28 Dec 2014 06:49:23 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Kusum",
"Amlan",
""
],
[
"Neamtiu",
"Iulian",
""
],
[
"Gupta",
"Rajiv",
""
]
] | TITLE: Adapting Graph Application Performance via Alternate Data Structure
Representation
ABSTRACT: Graph processing is used extensively in areas from social networking mining
to web indexing. We demonstrate that the performance and dependability of such
applications critically hinges on the graph data structure used, because a
fixed, compile-time choice of data structure can lead to poor performance or
applications unable to complete. To address this problem, we introduce an
approach that helps programmers transform regular, off-the-shelf graph
applications into adaptive, more dependable applications where adaptations are
performed via runtime selection from alternate data structure representations.
Using our approach, applications dynamically adapt to the input graph's
characteristics and changes in available memory so they continue to run when
faced with adverse conditions such as low memory. Experiments with graph
algorithms on real-world (e.g., Wikipedia metadata, Gnutella topology) and
synthetic graph datasets show that our adaptive applications run to completion
with lower execution time and/or memory utilization in comparison to their
non-adaptive versions.
|
1412.8341 | Pavel H\'ala | Pavel H\'ala | Spectral classification using convolutional neural networks | 71 pages, 50 figures, Master's thesis, Masaryk University | null | null | null | cs.CV astro-ph.IM cs.NE | http://creativecommons.org/licenses/by/3.0/ | There is a great need for accurate and autonomous spectral classification
methods in astrophysics. This thesis is about training a convolutional neural
network (ConvNet) to recognize an object class (quasar, star or galaxy) from
one-dimension spectra only. Author developed several scripts and C programs for
datasets preparation, preprocessing and postprocessing of the data. EBLearn
library (developed by Pierre Sermanet and Yann LeCun) was used to create
ConvNets. Application on dataset of more than 60000 spectra yielded success
rate of nearly 95%. This thesis conclusively proved great potential of
convolutional neural networks and deep learning methods in astrophysics.
| [
{
"version": "v1",
"created": "Mon, 29 Dec 2014 13:47:06 GMT"
}
] | 2014-12-30T00:00:00 | [
[
"Hála",
"Pavel",
""
]
] | TITLE: Spectral classification using convolutional neural networks
ABSTRACT: There is a great need for accurate and autonomous spectral classification
methods in astrophysics. This thesis is about training a convolutional neural
network (ConvNet) to recognize an object class (quasar, star or galaxy) from
one-dimension spectra only. Author developed several scripts and C programs for
datasets preparation, preprocessing and postprocessing of the data. EBLearn
library (developed by Pierre Sermanet and Yann LeCun) was used to create
ConvNets. Application on dataset of more than 60000 spectra yielded success
rate of nearly 95%. This thesis conclusively proved great potential of
convolutional neural networks and deep learning methods in astrophysics.
|
1412.7584 | Zhanglong Ji | Zhanglong Ji, Zachary C. Lipton, Charles Elkan | Differential Privacy and Machine Learning: a Survey and Review | null | null | null | null | cs.LG cs.CR cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The objective of machine learning is to extract useful information from data,
while privacy is preserved by concealing information. Thus it seems hard to
reconcile these competing interests. However, they frequently must be balanced
when mining sensitive data. For example, medical research represents an
important application where it is necessary both to extract useful information
and protect patient privacy. One way to resolve the conflict is to extract
general characteristics of whole populations without disclosing the private
information of individuals.
In this paper, we consider differential privacy, one of the most popular and
powerful definitions of privacy. We explore the interplay between machine
learning and differential privacy, namely privacy-preserving machine learning
algorithms and learning-based data release mechanisms. We also describe some
theoretical results that address what can be learned differentially privately
and upper bounds of loss functions for differentially private algorithms.
Finally, we present some open questions, including how to incorporate public
data, how to deal with missing data in private datasets, and whether, as the
number of observed samples grows arbitrarily large, differentially private
machine learning algorithms can be achieved at no cost to utility as compared
to corresponding non-differentially private algorithms.
| [
{
"version": "v1",
"created": "Wed, 24 Dec 2014 01:51:06 GMT"
}
] | 2014-12-25T00:00:00 | [
[
"Ji",
"Zhanglong",
""
],
[
"Lipton",
"Zachary C.",
""
],
[
"Elkan",
"Charles",
""
]
] | TITLE: Differential Privacy and Machine Learning: a Survey and Review
ABSTRACT: The objective of machine learning is to extract useful information from data,
while privacy is preserved by concealing information. Thus it seems hard to
reconcile these competing interests. However, they frequently must be balanced
when mining sensitive data. For example, medical research represents an
important application where it is necessary both to extract useful information
and protect patient privacy. One way to resolve the conflict is to extract
general characteristics of whole populations without disclosing the private
information of individuals.
In this paper, we consider differential privacy, one of the most popular and
powerful definitions of privacy. We explore the interplay between machine
learning and differential privacy, namely privacy-preserving machine learning
algorithms and learning-based data release mechanisms. We also describe some
theoretical results that address what can be learned differentially privately
and upper bounds of loss functions for differentially private algorithms.
Finally, we present some open questions, including how to incorporate public
data, how to deal with missing data in private datasets, and whether, as the
number of observed samples grows arbitrarily large, differentially private
machine learning algorithms can be achieved at no cost to utility as compared
to corresponding non-differentially private algorithms.
|
1412.6821 | Roland Kwitt | Jan Reininghaus, Stefan Huber, Ulrich Bauer, Roland Kwitt | A Stable Multi-Scale Kernel for Topological Machine Learning | null | null | null | null | stat.ML cs.CV cs.LG math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Topological data analysis offers a rich source of valuable information to
study vision problems. Yet, so far we lack a theoretically sound connection to
popular kernel-based learning techniques, such as kernel SVMs or kernel PCA. In
this work, we establish such a connection by designing a multi-scale kernel for
persistence diagrams, a stable summary representation of topological features
in data. We show that this kernel is positive definite and prove its stability
with respect to the 1-Wasserstein distance. Experiments on two benchmark
datasets for 3D shape classification/retrieval and texture recognition show
considerable performance gains of the proposed method compared to an
alternative approach that is based on the recently introduced persistence
landscapes.
| [
{
"version": "v1",
"created": "Sun, 21 Dec 2014 19:17:08 GMT"
}
] | 2014-12-24T00:00:00 | [
[
"Reininghaus",
"Jan",
""
],
[
"Huber",
"Stefan",
""
],
[
"Bauer",
"Ulrich",
""
],
[
"Kwitt",
"Roland",
""
]
] | TITLE: A Stable Multi-Scale Kernel for Topological Machine Learning
ABSTRACT: Topological data analysis offers a rich source of valuable information to
study vision problems. Yet, so far we lack a theoretically sound connection to
popular kernel-based learning techniques, such as kernel SVMs or kernel PCA. In
this work, we establish such a connection by designing a multi-scale kernel for
persistence diagrams, a stable summary representation of topological features
in data. We show that this kernel is positive definite and prove its stability
with respect to the 1-Wasserstein distance. Experiments on two benchmark
datasets for 3D shape classification/retrieval and texture recognition show
considerable performance gains of the proposed method compared to an
alternative approach that is based on the recently introduced persistence
landscapes.
|
1209.3686 | Barzan Mozafari | Barzan Mozafari, Purnamrita Sarkar, Michael J. Franklin, Michael I.
Jordan, Samuel Madden | Active Learning for Crowd-Sourced Databases | A shorter version of this manuscript has been published in
Proceedings of Very Large Data Bases 2015, entitled "Scaling Up
Crowd-Sourcing to Very Large Datasets: A Case for Active Learning" | null | null | null | cs.LG cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Crowd-sourcing has become a popular means of acquiring labeled data for a
wide variety of tasks where humans are more accurate than computers, e.g.,
labeling images, matching objects, or analyzing sentiment. However, relying
solely on the crowd is often impractical even for data sets with thousands of
items, due to time and cost constraints of acquiring human input (which cost
pennies and minutes per label). In this paper, we propose algorithms for
integrating machine learning into crowd-sourced databases, with the goal of
allowing crowd-sourcing applications to scale, i.e., to handle larger datasets
at lower costs. The key observation is that, in many of the above tasks, humans
and machine learning algorithms can be complementary, as humans are often more
accurate but slow and expensive, while algorithms are usually less accurate,
but faster and cheaper.
Based on this observation, we present two new active learning algorithms to
combine humans and algorithms together in a crowd-sourced database. Our
algorithms are based on the theory of non-parametric bootstrap, which makes our
results applicable to a broad class of machine learning models. Our results, on
three real-life datasets collected with Amazon's Mechanical Turk, and on 15
well-known UCI data sets, show that our methods on average ask humans to label
one to two orders of magnitude fewer items to achieve the same accuracy as a
baseline that labels random images, and two to eight times fewer questions than
previous active learning schemes.
| [
{
"version": "v1",
"created": "Mon, 17 Sep 2012 15:21:06 GMT"
},
{
"version": "v2",
"created": "Mon, 3 Dec 2012 15:45:55 GMT"
},
{
"version": "v3",
"created": "Thu, 13 Dec 2012 18:20:04 GMT"
},
{
"version": "v4",
"created": "Sat, 20 Dec 2014 08:56:15 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Mozafari",
"Barzan",
""
],
[
"Sarkar",
"Purnamrita",
""
],
[
"Franklin",
"Michael J.",
""
],
[
"Jordan",
"Michael I.",
""
],
[
"Madden",
"Samuel",
""
]
] | TITLE: Active Learning for Crowd-Sourced Databases
ABSTRACT: Crowd-sourcing has become a popular means of acquiring labeled data for a
wide variety of tasks where humans are more accurate than computers, e.g.,
labeling images, matching objects, or analyzing sentiment. However, relying
solely on the crowd is often impractical even for data sets with thousands of
items, due to time and cost constraints of acquiring human input (which cost
pennies and minutes per label). In this paper, we propose algorithms for
integrating machine learning into crowd-sourced databases, with the goal of
allowing crowd-sourcing applications to scale, i.e., to handle larger datasets
at lower costs. The key observation is that, in many of the above tasks, humans
and machine learning algorithms can be complementary, as humans are often more
accurate but slow and expensive, while algorithms are usually less accurate,
but faster and cheaper.
Based on this observation, we present two new active learning algorithms to
combine humans and algorithms together in a crowd-sourced database. Our
algorithms are based on the theory of non-parametric bootstrap, which makes our
results applicable to a broad class of machine learning models. Our results, on
three real-life datasets collected with Amazon's Mechanical Turk, and on 15
well-known UCI data sets, show that our methods on average ask humans to label
one to two orders of magnitude fewer items to achieve the same accuracy as a
baseline that labels random images, and two to eight times fewer questions than
previous active learning schemes.
|
1304.5409 | Carsten Gottschlich | Carsten Gottschlich and Stephan Huckemann | Separating the Real from the Synthetic: Minutiae Histograms as
Fingerprints of Fingerprints | null | IET Biometrics, vol. 3, no. 4, pp. 291-301, Dec. 2014 | 10.1049/iet-bmt.2013.0065 | null | cs.CV cs.AI cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this study we show that by the current state-of-the-art synthetically
generated fingerprints can easily be discriminated from real fingerprints. We
propose a method based on second order extended minutiae histograms (MHs) which
can distinguish between real and synthetic prints with very high accuracy. MHs
provide a fixed-length feature vector for a fingerprint which are invariant
under rotation and translation. This 'test of realness' can be applied to
synthetic fingerprints produced by any method. In this work, tests are
conducted on the 12 publicly available databases of FVC2000, FVC2002 and
FVC2004 which are well established benchmarks for evaluating the performance of
fingerprint recognition algorithms; 3 of these 12 databases consist of
artificial fingerprints generated by the SFinGe software. Additionally, we
evaluate the discriminative performance on a database of synthetic fingerprints
generated by the software of Bicz versus real fingerprint images. We conclude
with suggestions for the improvement of synthetic fingerprint generation.
| [
{
"version": "v1",
"created": "Fri, 19 Apr 2013 13:21:13 GMT"
},
{
"version": "v2",
"created": "Thu, 30 Jan 2014 11:28:08 GMT"
},
{
"version": "v3",
"created": "Wed, 15 Oct 2014 14:04:46 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Gottschlich",
"Carsten",
""
],
[
"Huckemann",
"Stephan",
""
]
] | TITLE: Separating the Real from the Synthetic: Minutiae Histograms as
Fingerprints of Fingerprints
ABSTRACT: In this study we show that by the current state-of-the-art synthetically
generated fingerprints can easily be discriminated from real fingerprints. We
propose a method based on second order extended minutiae histograms (MHs) which
can distinguish between real and synthetic prints with very high accuracy. MHs
provide a fixed-length feature vector for a fingerprint which are invariant
under rotation and translation. This 'test of realness' can be applied to
synthetic fingerprints produced by any method. In this work, tests are
conducted on the 12 publicly available databases of FVC2000, FVC2002 and
FVC2004 which are well established benchmarks for evaluating the performance of
fingerprint recognition algorithms; 3 of these 12 databases consist of
artificial fingerprints generated by the SFinGe software. Additionally, we
evaluate the discriminative performance on a database of synthetic fingerprints
generated by the software of Bicz versus real fingerprint images. We conclude
with suggestions for the improvement of synthetic fingerprint generation.
|
1404.3840 | Chaochao Lu | Chaochao Lu, Xiaoou Tang | Surpassing Human-Level Face Verification Performance on LFW with
GaussianFace | Appearing in Proceedings of the 29th AAAI Conference on Artificial
Intelligence (AAAI-15), Oral Presentation | null | null | null | cs.CV cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Face verification remains a challenging problem in very complex conditions
with large variations such as pose, illumination, expression, and occlusions.
This problem is exacerbated when we rely unrealistically on a single training
data source, which is often insufficient to cover the intrinsically complex
face variations. This paper proposes a principled multi-task learning approach
based on Discriminative Gaussian Process Latent Variable Model, named
GaussianFace, to enrich the diversity of training data. In comparison to
existing methods, our model exploits additional data from multiple
source-domains to improve the generalization performance of face verification
in an unknown target-domain. Importantly, our model can adapt automatically to
complex data distributions, and therefore can well capture complex face
variations inherent in multiple sources. Extensive experiments demonstrate the
effectiveness of the proposed model in learning from diverse data sources and
generalize to unseen domain. Specifically, the accuracy of our algorithm
achieves an impressive accuracy rate of 98.52% on the well-known and
challenging Labeled Faces in the Wild (LFW) benchmark. For the first time, the
human-level performance in face verification (97.53%) on LFW is surpassed.
| [
{
"version": "v1",
"created": "Tue, 15 Apr 2014 07:51:23 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Jun 2014 14:37:38 GMT"
},
{
"version": "v3",
"created": "Sat, 20 Dec 2014 03:37:36 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Lu",
"Chaochao",
""
],
[
"Tang",
"Xiaoou",
""
]
] | TITLE: Surpassing Human-Level Face Verification Performance on LFW with
GaussianFace
ABSTRACT: Face verification remains a challenging problem in very complex conditions
with large variations such as pose, illumination, expression, and occlusions.
This problem is exacerbated when we rely unrealistically on a single training
data source, which is often insufficient to cover the intrinsically complex
face variations. This paper proposes a principled multi-task learning approach
based on Discriminative Gaussian Process Latent Variable Model, named
GaussianFace, to enrich the diversity of training data. In comparison to
existing methods, our model exploits additional data from multiple
source-domains to improve the generalization performance of face verification
in an unknown target-domain. Importantly, our model can adapt automatically to
complex data distributions, and therefore can well capture complex face
variations inherent in multiple sources. Extensive experiments demonstrate the
effectiveness of the proposed model in learning from diverse data sources and
generalize to unseen domain. Specifically, the accuracy of our algorithm
achieves an impressive accuracy rate of 98.52% on the well-known and
challenging Labeled Faces in the Wild (LFW) benchmark. For the first time, the
human-level performance in face verification (97.53%) on LFW is surpassed.
|
1412.6570 | Changchun Zhang | Robert C. Qiu | The Foundation of Big Data: Experiments, Formulation, and Applications | null | null | null | null | cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The central theme of this talk is to promote the non-asymptotic statistical
viewpoint in the context of massive datasets. The classical viewpoint breaks
down when the data size becomes large.
| [
{
"version": "v1",
"created": "Sat, 20 Dec 2014 01:14:55 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Qiu",
"Robert C.",
""
]
] | TITLE: The Foundation of Big Data: Experiments, Formulation, and Applications
ABSTRACT: The central theme of this talk is to promote the non-asymptotic statistical
viewpoint in the context of massive datasets. The classical viewpoint breaks
down when the data size becomes large.
|
1412.6734 | Aviv Tamar | Aviv Tamar, Panos Toulis, Shie Mannor, Edoardo M. Airoldi | Implicit Temporal Differences | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In reinforcement learning, the TD($\lambda$) algorithm is a fundamental
policy evaluation method with an efficient online implementation that is
suitable for large-scale problems. One practical drawback of TD($\lambda$) is
its sensitivity to the choice of the step-size. It is an empirically well-known
fact that a large step-size leads to fast convergence, at the cost of higher
variance and risk of instability. In this work, we introduce the implicit
TD($\lambda$) algorithm which has the same function and computational cost as
TD($\lambda$), but is significantly more stable. We provide a theoretical
explanation of this stability and an empirical evaluation of implicit
TD($\lambda$) on typical benchmark tasks. Our results show that implicit
TD($\lambda$) outperforms standard TD($\lambda$) and a state-of-the-art method
that automatically tunes the step-size, and thus shows promise for wide
applicability.
| [
{
"version": "v1",
"created": "Sun, 21 Dec 2014 06:53:15 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Tamar",
"Aviv",
""
],
[
"Toulis",
"Panos",
""
],
[
"Mannor",
"Shie",
""
],
[
"Airoldi",
"Edoardo M.",
""
]
] | TITLE: Implicit Temporal Differences
ABSTRACT: In reinforcement learning, the TD($\lambda$) algorithm is a fundamental
policy evaluation method with an efficient online implementation that is
suitable for large-scale problems. One practical drawback of TD($\lambda$) is
its sensitivity to the choice of the step-size. It is an empirically well-known
fact that a large step-size leads to fast convergence, at the cost of higher
variance and risk of instability. In this work, we introduce the implicit
TD($\lambda$) algorithm which has the same function and computational cost as
TD($\lambda$), but is significantly more stable. We provide a theoretical
explanation of this stability and an empirical evaluation of implicit
TD($\lambda$) on typical benchmark tasks. Our results show that implicit
TD($\lambda$) outperforms standard TD($\lambda$) and a state-of-the-art method
that automatically tunes the step-size, and thus shows promise for wide
applicability.
|
1412.6791 | Anoop Katti | Anoop Katti, Anurag Mittal | Mixture of Parts Revisited: Expressive Part Interactions for Pose
Estimation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Part-based models with restrictive tree-structured interactions for the Human
Pose Estimation problem, leaves many part interactions unhandled. Two of the
most common and strong manifestations of such unhandled interactions are
self-occlusion among the parts and the confusion in the localization of the
non-adjacent symmetric parts. By handling the self-occlusion in a data
efficient manner, we improve the performance of the basic Mixture of Parts
model by a large margin, especially on uncommon poses. Through addressing the
confusion in the symmetric limb localization using a combination of two
complementing trees, we improve the performance on all the parts by atmost
doubling the running time. Finally, we show that the combination of the two
solutions improves the results. We report results that are equivalent to the
state-of-the-art on two standard datasets. Because of maintaining the
tree-structured interactions and only part-level modeling of the base Mixture
of Parts model, this is achieved in time that is much less than the best
performing part-based model.
| [
{
"version": "v1",
"created": "Sun, 21 Dec 2014 14:48:41 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Katti",
"Anoop",
""
],
[
"Mittal",
"Anurag",
""
]
] | TITLE: Mixture of Parts Revisited: Expressive Part Interactions for Pose
Estimation
ABSTRACT: Part-based models with restrictive tree-structured interactions for the Human
Pose Estimation problem, leaves many part interactions unhandled. Two of the
most common and strong manifestations of such unhandled interactions are
self-occlusion among the parts and the confusion in the localization of the
non-adjacent symmetric parts. By handling the self-occlusion in a data
efficient manner, we improve the performance of the basic Mixture of Parts
model by a large margin, especially on uncommon poses. Through addressing the
confusion in the symmetric limb localization using a combination of two
complementing trees, we improve the performance on all the parts by atmost
doubling the running time. Finally, we show that the combination of the two
solutions improves the results. We report results that are equivalent to the
state-of-the-art on two standard datasets. Because of maintaining the
tree-structured interactions and only part-level modeling of the base Mixture
of Parts model, this is achieved in time that is much less than the best
performing part-based model.
|
1412.6883 | Mahdi Nasrullah Al-Ameen | Mahdi Nasrullah Al-Ameen and Matthew Wright | iPersea : The Improved Persea with Sybil Detection Mechanism | 10 pages | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | P2P systems are highly susceptible to Sybil attacks, in which an attacker
creates a large number of identities and uses them to control a substantial
fraction of the system. Persea is the most recent approach towards designing a
social network based Sybil-resistant DHT. Unlike prior Sybil-resistant P2P
systems based on social networks, Persea does not rely on two key assumptions:
(i) that the social network is fast mixing, and (ii) that there is a small
ratio of attack edges to honest peers. Both assumptions have been shown to be
unreliable in real social networks. The hierarchical distribution of node IDs
in Persea confines a large attacker botnet to a considerably smaller region of
the ID space than in a normal P2P system and its replication mechanism lets a
peer to retrieve the desired results even if a given region is occupied by
attackers. However, Persea system suffers from certain limitations, since it
cannot handle the scenario, where the malicious target returns an incorrect
result instead of just ignoring the lookup request. In this paper, we address
this major limitation of Persea through a Sybil detection mechanism built on
top of Persea system, which accommodates inspection lookup, a specially
designed lookup scheme to detect the Sybil nodes based on their responses to
the lookup query. We design a scheme to filter those detected Sybils to ensure
the participation of honest nodes on the lookup path during regular DHT lookup.
Since the malicious nodes are opt-out from the lookup path in our system, they
cannot return any incorrect result during regular lookup. We evaluate our
system in simulations with social network datasets and the results show that
catster, the largest network in our simulation with 149700 nodes and 5449275
edges, gains 100% lookup success rate, even when the number of attack edges is
equal to the number of benign peers in the network.
| [
{
"version": "v1",
"created": "Mon, 22 Dec 2014 06:25:49 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Al-Ameen",
"Mahdi Nasrullah",
""
],
[
"Wright",
"Matthew",
""
]
] | TITLE: iPersea : The Improved Persea with Sybil Detection Mechanism
ABSTRACT: P2P systems are highly susceptible to Sybil attacks, in which an attacker
creates a large number of identities and uses them to control a substantial
fraction of the system. Persea is the most recent approach towards designing a
social network based Sybil-resistant DHT. Unlike prior Sybil-resistant P2P
systems based on social networks, Persea does not rely on two key assumptions:
(i) that the social network is fast mixing, and (ii) that there is a small
ratio of attack edges to honest peers. Both assumptions have been shown to be
unreliable in real social networks. The hierarchical distribution of node IDs
in Persea confines a large attacker botnet to a considerably smaller region of
the ID space than in a normal P2P system and its replication mechanism lets a
peer to retrieve the desired results even if a given region is occupied by
attackers. However, Persea system suffers from certain limitations, since it
cannot handle the scenario, where the malicious target returns an incorrect
result instead of just ignoring the lookup request. In this paper, we address
this major limitation of Persea through a Sybil detection mechanism built on
top of Persea system, which accommodates inspection lookup, a specially
designed lookup scheme to detect the Sybil nodes based on their responses to
the lookup query. We design a scheme to filter those detected Sybils to ensure
the participation of honest nodes on the lookup path during regular DHT lookup.
Since the malicious nodes are opt-out from the lookup path in our system, they
cannot return any incorrect result during regular lookup. We evaluate our
system in simulations with social network datasets and the results show that
catster, the largest network in our simulation with 149700 nodes and 5449275
edges, gains 100% lookup success rate, even when the number of attack edges is
equal to the number of benign peers in the network.
|
1412.6986 | Tarek Abdelrahman | Tianyi David Han, Tarek S. Abdelrahman | Automatic Tuning of Local Memory Use on GPGPUs | Part of ADAPT Workshop proceedings, 2015 (arXiv:1412.2347) | null | null | ADAPT/2015/04 | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The use of local memory is important to improve the performance of OpenCL
programs. However, its use may not always benefit performance, depending on
various application characteristics, and there is no simple heuristic for
deciding when to use it. We develop a machine learning model to decide if the
optimization is beneficial or not. We train the model with millions of
synthetic benchmarks and show that it can predict if the optimization should be
applied for a single array, in both synthetic and real benchmarks, with high
accuracy.
| [
{
"version": "v1",
"created": "Mon, 22 Dec 2014 14:06:37 GMT"
}
] | 2014-12-23T00:00:00 | [
[
"Han",
"Tianyi David",
""
],
[
"Abdelrahman",
"Tarek S.",
""
]
] | TITLE: Automatic Tuning of Local Memory Use on GPGPUs
ABSTRACT: The use of local memory is important to improve the performance of OpenCL
programs. However, its use may not always benefit performance, depending on
various application characteristics, and there is no simple heuristic for
deciding when to use it. We develop a machine learning model to decide if the
optimization is beneficial or not. We train the model with millions of
synthetic benchmarks and show that it can predict if the optimization should be
applied for a single array, in both synthetic and real benchmarks, with high
accuracy.
|
1412.6124 | Jianyu Wang | Jianyu Wang and Alan Yuille | Semantic Part Segmentation using Compositional Model combining Shape and
Appearance | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study the problem of semantic part segmentation for
animals. This is more challenging than standard object detection, object
segmentation and pose estimation tasks because semantic parts of animals often
have similar appearance and highly varying shapes. To tackle these challenges,
we build a mixture of compositional models to represent the object boundary and
the boundaries of semantic parts. And we incorporate edge, appearance, and
semantic part cues into the compositional model. Given part-level segmentation
annotation, we develop a novel algorithm to learn a mixture of compositional
models under various poses and viewpoints for certain animal classes.
Furthermore, a linear complexity algorithm is offered for efficient inference
of the compositional model using dynamic programming. We evaluate our method
for horse and cow using a newly annotated dataset on Pascal VOC 2010 which has
pixelwise part labels. Experimental results demonstrate the effectiveness of
our method.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 21:27:38 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"Wang",
"Jianyu",
""
],
[
"Yuille",
"Alan",
""
]
] | TITLE: Semantic Part Segmentation using Compositional Model combining Shape and
Appearance
ABSTRACT: In this paper, we study the problem of semantic part segmentation for
animals. This is more challenging than standard object detection, object
segmentation and pose estimation tasks because semantic parts of animals often
have similar appearance and highly varying shapes. To tackle these challenges,
we build a mixture of compositional models to represent the object boundary and
the boundaries of semantic parts. And we incorporate edge, appearance, and
semantic part cues into the compositional model. Given part-level segmentation
annotation, we develop a novel algorithm to learn a mixture of compositional
models under various poses and viewpoints for certain animal classes.
Furthermore, a linear complexity algorithm is offered for efficient inference
of the compositional model using dynamic programming. We evaluate our method
for horse and cow using a newly annotated dataset on Pascal VOC 2010 which has
pixelwise part labels. Experimental results demonstrate the effectiveness of
our method.
|
1412.6154 | Ana Romero | Ana Romero, Julio Rubio, Francis Sergeraert | Effective persistent homology of digital images | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, three Computational Topology methods (namely effective
homology, persistent homology and discrete vector fields) are mixed together to
produce algorithms for homological digital image processing. The algorithms
have been implemented as extensions of the Kenzo system and have shown a good
performance when applied on some actual images extracted from a public dataset.
| [
{
"version": "v1",
"created": "Mon, 6 Oct 2014 11:45:07 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"Romero",
"Ana",
""
],
[
"Rubio",
"Julio",
""
],
[
"Sergeraert",
"Francis",
""
]
] | TITLE: Effective persistent homology of digital images
ABSTRACT: In this paper, three Computational Topology methods (namely effective
homology, persistent homology and discrete vector fields) are mixed together to
produce algorithms for homological digital image processing. The algorithms
have been implemented as extensions of the Kenzo system and have shown a good
performance when applied on some actual images extracted from a public dataset.
|
1412.6170 | Francesco Lettich | Francesco Lettich, Salvatore Orlando and Claudio Silvestri | Manycore processing of repeated k-NN queries over massive moving objects
observations | null | null | null | null | cs.DC cs.DB cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ability to timely process significant amounts of continuously updated
spatial data is mandatory for an increasing number of applications. In this
paper we focus on a specific data-intensive problem concerning the repeated
processing of huge amounts of k nearest neighbours (k-NN) queries over massive
sets of moving objects, where the spatial extents of queries and the position
of objects are continuously modified over time. In particular, we propose a
novel hybrid CPU/GPU pipeline that significantly accelerate query processing
thanks to a combination of ad-hoc data structures and non-trivial memory access
patterns. To the best of our knowledge this is the first work that exploits
GPUs to efficiently solve repeated k-NN queries over massive sets of
continuously moving objects, even characterized by highly skewed spatial
distributions. In comparison with state-of-the-art sequential CPU-based
implementations, our method highlights significant speedups in the order of
10x-20x, depending on the datasets, even when considering cheap GPUs.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 22:43:28 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"Lettich",
"Francesco",
""
],
[
"Orlando",
"Salvatore",
""
],
[
"Silvestri",
"Claudio",
""
]
] | TITLE: Manycore processing of repeated k-NN queries over massive moving objects
observations
ABSTRACT: The ability to timely process significant amounts of continuously updated
spatial data is mandatory for an increasing number of applications. In this
paper we focus on a specific data-intensive problem concerning the repeated
processing of huge amounts of k nearest neighbours (k-NN) queries over massive
sets of moving objects, where the spatial extents of queries and the position
of objects are continuously modified over time. In particular, we propose a
novel hybrid CPU/GPU pipeline that significantly accelerate query processing
thanks to a combination of ad-hoc data structures and non-trivial memory access
patterns. To the best of our knowledge this is the first work that exploits
GPUs to efficiently solve repeated k-NN queries over massive sets of
continuously moving objects, even characterized by highly skewed spatial
distributions. In comparison with state-of-the-art sequential CPU-based
implementations, our method highlights significant speedups in the order of
10x-20x, depending on the datasets, even when considering cheap GPUs.
|
1412.6257 | Alexander Kalmanovich | Alexander Kalmanovich and Gal Chechik | Gradual training of deep denoising auto encoders | null | null | null | null | cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Stacked denoising auto encoders (DAEs) are well known to learn useful deep
representations, which can be used to improve supervised training by
initializing a deep network. We investigate a training scheme of a deep DAE,
where DAE layers are gradually added and keep adapting as additional layers are
added. We show that in the regime of mid-sized datasets, this gradual training
provides a small but consistent improvement over stacked training in both
reconstruction quality and classification error over stacked training on MNIST
and CIFAR datasets.
| [
{
"version": "v1",
"created": "Fri, 19 Dec 2014 09:30:33 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"Kalmanovich",
"Alexander",
""
],
[
"Chechik",
"Gal",
""
]
] | TITLE: Gradual training of deep denoising auto encoders
ABSTRACT: Stacked denoising auto encoders (DAEs) are well known to learn useful deep
representations, which can be used to improve supervised training by
initializing a deep network. We investigate a training scheme of a deep DAE,
where DAE layers are gradually added and keep adapting as additional layers are
added. We show that in the regime of mid-sized datasets, this gradual training
provides a small but consistent improvement over stacked training in both
reconstruction quality and classification error over stacked training on MNIST
and CIFAR datasets.
|
1412.6264 | Taraka Rama Kasicheyanula | Taraka Rama K | Supertagging: Introduction, learning, and application | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/3.0/ | Supertagging is an approach originally developed by Bangalore and Joshi
(1999) to improve the parsing efficiency. In the beginning, the scholars used
small training datasets and somewhat na\"ive smoothing techniques to learn the
probability distributions of supertags. Since its inception, the applicability
of Supertags has been explored for TAG (tree-adjoining grammar) formalism as
well as other related yet, different formalisms such as CCG. This article will
try to summarize the various chapters, relevant to statistical parsing, from
the most recent edited book volume (Bangalore and Joshi, 2010). The chapters
were selected so as to blend the learning of supertags, its integration into
full-scale parsing, and in semantic parsing.
| [
{
"version": "v1",
"created": "Fri, 19 Dec 2014 09:53:57 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"K",
"Taraka Rama",
""
]
] | TITLE: Supertagging: Introduction, learning, and application
ABSTRACT: Supertagging is an approach originally developed by Bangalore and Joshi
(1999) to improve the parsing efficiency. In the beginning, the scholars used
small training datasets and somewhat na\"ive smoothing techniques to learn the
probability distributions of supertags. Since its inception, the applicability
of Supertags has been explored for TAG (tree-adjoining grammar) formalism as
well as other related yet, different formalisms such as CCG. This article will
try to summarize the various chapters, relevant to statistical parsing, from
the most recent edited book volume (Bangalore and Joshi, 2010). The chapters
were selected so as to blend the learning of supertags, its integration into
full-scale parsing, and in semantic parsing.
|
1412.6402 | Pierre de Buyl | Rebecca R. Murphy, Sophie E. Jackson, David Klenerman | pyFRET: A Python Library for Single Molecule Fluorescence Data Analysis | Part of the Proceedings of the 7th European Conference on Python in
Science (EuroSciPy 2014), Pierre de Buyl and Nelle Varoquaux editors, (2014) | null | null | euroscipy-proceedings2014-10 | cs.CE physics.bio-ph q-bio.BM | http://creativecommons.org/licenses/by/3.0/ | Single molecule F\"orster resonance energy transfer (smFRET) is a powerful
experimental technique for studying the properties of individual biological
molecules in solution. However, as adoption of smFRET techniques becomes more
widespread, the lack of available software, whether open source or commercial,
for data analysis, is becoming a significant issue. Here, we present pyFRET, an
open source Python package for the analysis of data from single-molecule
fluorescence experiments from freely diffusing biomolecules. The package
provides methods for the complete analysis of a smFRET dataset, from burst
selection and denoising, through data visualisation and model fitting. We
provide support for both continuous excitation and alternating laser excitation
(ALEX) data analysis. pyFRET is available as a package downloadable from the
Python Package Index (PyPI) under the open source three-clause BSD licence,
together with links to extensive documentation and tutorials, including example
usage and test data. Additional documentation including tutorials is hosted
independently on ReadTheDocs. The code is available from the free hosting site
Bitbucket. Through distribution of this software, we hope to lower the barrier
for the adoption of smFRET experiments by other research groups and we
encourage others to contribute modules for specific analysis needs.
| [
{
"version": "v1",
"created": "Fri, 19 Dec 2014 16:00:31 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"Murphy",
"Rebecca R.",
""
],
[
"Jackson",
"Sophie E.",
""
],
[
"Klenerman",
"David",
""
]
] | TITLE: pyFRET: A Python Library for Single Molecule Fluorescence Data Analysis
ABSTRACT: Single molecule F\"orster resonance energy transfer (smFRET) is a powerful
experimental technique for studying the properties of individual biological
molecules in solution. However, as adoption of smFRET techniques becomes more
widespread, the lack of available software, whether open source or commercial,
for data analysis, is becoming a significant issue. Here, we present pyFRET, an
open source Python package for the analysis of data from single-molecule
fluorescence experiments from freely diffusing biomolecules. The package
provides methods for the complete analysis of a smFRET dataset, from burst
selection and denoising, through data visualisation and model fitting. We
provide support for both continuous excitation and alternating laser excitation
(ALEX) data analysis. pyFRET is available as a package downloadable from the
Python Package Index (PyPI) under the open source three-clause BSD licence,
together with links to extensive documentation and tutorials, including example
usage and test data. Additional documentation including tutorials is hosted
independently on ReadTheDocs. The code is available from the free hosting site
Bitbucket. Through distribution of this software, we hope to lower the barrier
for the adoption of smFRET experiments by other research groups and we
encourage others to contribute modules for specific analysis needs.
|
1412.6493 | Zichao Yang | Zichao Yang and Alexander J. Smola and Le Song and Andrew Gordon
Wilson | A la Carte - Learning Fast Kernels | null | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Kernel methods have great promise for learning rich statistical
representations of large modern datasets. However, compared to neural networks,
kernel methods have been perceived as lacking in scalability and flexibility.
We introduce a family of fast, flexible, lightly parametrized and general
purpose kernel learning methods, derived from Fastfood basis function
expansions. We provide mechanisms to learn the properties of groups of spectral
frequencies in these expansions, which require only O(mlogd) time and O(m)
memory, for m basis functions and d input dimensions. We show that the proposed
methods can learn a wide class of kernels, outperforming the alternatives in
accuracy, speed, and memory consumption.
| [
{
"version": "v1",
"created": "Fri, 19 Dec 2014 19:27:21 GMT"
}
] | 2014-12-22T00:00:00 | [
[
"Yang",
"Zichao",
""
],
[
"Smola",
"Alexander J.",
""
],
[
"Song",
"Le",
""
],
[
"Wilson",
"Andrew Gordon",
""
]
] | TITLE: A la Carte - Learning Fast Kernels
ABSTRACT: Kernel methods have great promise for learning rich statistical
representations of large modern datasets. However, compared to neural networks,
kernel methods have been perceived as lacking in scalability and flexibility.
We introduce a family of fast, flexible, lightly parametrized and general
purpose kernel learning methods, derived from Fastfood basis function
expansions. We provide mechanisms to learn the properties of groups of spectral
frequencies in these expansions, which require only O(mlogd) time and O(m)
memory, for m basis functions and d input dimensions. We show that the proposed
methods can learn a wide class of kernels, outperforming the alternatives in
accuracy, speed, and memory consumption.
|
1303.1624 | Conrad Sanderson | Yongkang Wong, Mehrtash T. Harandi, Conrad Sanderson | On Robust Face Recognition via Sparse Encoding: the Good, the Bad, and
the Ugly | null | IET Biometrics, Vol. 3, No. 4, pp. 176-189, 2014 | 10.1049/iet-bmt.2013.0033 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the field of face recognition, Sparse Representation (SR) has received
considerable attention during the past few years. Most of the relevant
literature focuses on holistic descriptors in closed-set identification
applications. The underlying assumption in SR-based methods is that each class
in the gallery has sufficient samples and the query lies on the subspace
spanned by the gallery of the same class. Unfortunately, such assumption is
easily violated in the more challenging face verification scenario, where an
algorithm is required to determine if two faces (where one or both have not
been seen before) belong to the same person. In this paper, we first discuss
why previous attempts with SR might not be applicable to verification problems.
We then propose an alternative approach to face verification via SR.
Specifically, we propose to use explicit SR encoding on local image patches
rather than the entire face. The obtained sparse signals are pooled via
averaging to form multiple region descriptors, which are then concatenated to
form an overall face descriptor. Due to the deliberate loss spatial relations
within each region (caused by averaging), the resulting descriptor is robust to
misalignment & various image deformations. Within the proposed framework, we
evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder
Neural Network (SANN), and an implicit probabilistic technique based on
Gaussian Mixture Models. Thorough experiments on AR, FERET, exYaleB, BANCA and
ChokePoint datasets show that the proposed local SR approach obtains
considerably better and more robust performance than several previous
state-of-the-art holistic SR methods, in both verification and closed-set
identification problems. The experiments also show that l1-minimisation based
encoding has a considerably higher computational than the other techniques, but
leads to higher recognition rates.
| [
{
"version": "v1",
"created": "Thu, 7 Mar 2013 09:30:10 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Wong",
"Yongkang",
""
],
[
"Harandi",
"Mehrtash T.",
""
],
[
"Sanderson",
"Conrad",
""
]
] | TITLE: On Robust Face Recognition via Sparse Encoding: the Good, the Bad, and
the Ugly
ABSTRACT: In the field of face recognition, Sparse Representation (SR) has received
considerable attention during the past few years. Most of the relevant
literature focuses on holistic descriptors in closed-set identification
applications. The underlying assumption in SR-based methods is that each class
in the gallery has sufficient samples and the query lies on the subspace
spanned by the gallery of the same class. Unfortunately, such assumption is
easily violated in the more challenging face verification scenario, where an
algorithm is required to determine if two faces (where one or both have not
been seen before) belong to the same person. In this paper, we first discuss
why previous attempts with SR might not be applicable to verification problems.
We then propose an alternative approach to face verification via SR.
Specifically, we propose to use explicit SR encoding on local image patches
rather than the entire face. The obtained sparse signals are pooled via
averaging to form multiple region descriptors, which are then concatenated to
form an overall face descriptor. Due to the deliberate loss spatial relations
within each region (caused by averaging), the resulting descriptor is robust to
misalignment & various image deformations. Within the proposed framework, we
evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder
Neural Network (SANN), and an implicit probabilistic technique based on
Gaussian Mixture Models. Thorough experiments on AR, FERET, exYaleB, BANCA and
ChokePoint datasets show that the proposed local SR approach obtains
considerably better and more robust performance than several previous
state-of-the-art holistic SR methods, in both verification and closed-set
identification problems. The experiments also show that l1-minimisation based
encoding has a considerably higher computational than the other techniques, but
leads to higher recognition rates.
|
1412.3506 | Jose M. Alvarez | Jose M. Alvarez and Theo Gevers and Antonio M. Lopez | Road Detection by One-Class Color Classification: Dataset and
Experiments | 10 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Detecting traversable road areas ahead a moving vehicle is a key process for
modern autonomous driving systems. A common approach to road detection consists
of exploiting color features to classify pixels as road or background. These
algorithms reduce the effect of lighting variations and weather conditions by
exploiting the discriminant/invariant properties of different color
representations. Furthermore, the lack of labeled datasets has motivated the
development of algorithms performing on single images based on the assumption
that the bottom part of the image belongs to the road surface.
In this paper, we first introduce a dataset of road images taken at different
times and in different scenarios using an onboard camera. Then, we devise a
simple online algorithm and conduct an exhaustive evaluation of different
classifiers and the effect of using different color representation to
characterize pixels.
| [
{
"version": "v1",
"created": "Thu, 11 Dec 2014 00:31:37 GMT"
},
{
"version": "v2",
"created": "Thu, 18 Dec 2014 00:57:36 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Alvarez",
"Jose M.",
""
],
[
"Gevers",
"Theo",
""
],
[
"Lopez",
"Antonio M.",
""
]
] | TITLE: Road Detection by One-Class Color Classification: Dataset and
Experiments
ABSTRACT: Detecting traversable road areas ahead a moving vehicle is a key process for
modern autonomous driving systems. A common approach to road detection consists
of exploiting color features to classify pixels as road or background. These
algorithms reduce the effect of lighting variations and weather conditions by
exploiting the discriminant/invariant properties of different color
representations. Furthermore, the lack of labeled datasets has motivated the
development of algorithms performing on single images based on the assumption
that the bottom part of the image belongs to the road surface.
In this paper, we first introduce a dataset of road images taken at different
times and in different scenarios using an onboard camera. Then, we devise a
simple online algorithm and conduct an exhaustive evaluation of different
classifiers and the effect of using different color representation to
characterize pixels.
|
1412.4973 | Somnath Sikdar | Jan Dreier and Philipp Kuinke and Rafael Przybylski and Felix Reidl
and Peter Rossmanith and Somnath Sikdar | Overlapping Communities in Social Networks | null | null | null | null | cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Complex networks can be typically broken down into groups or modules.
Discovering this "community structure" is an important step in studying the
large-scale structure of networks. Many algorithms have been proposed for
community detection and benchmarks have been created to evaluate their
performance. Typically algorithms for community detection either partition the
graph (non-overlapping communities) or find node covers (overlapping
communities).
In this paper, we propose a particularly simple semi-supervised learning
algorithm for finding out communities. In essence, given the community
information of a small number of "seed nodes", the method uses random walks
from the seed nodes to uncover the community information of the whole network.
The algorithm runs in time $O(k \cdot m \cdot \log n)$, where $m$ is the number
of edges; $n$ the number of links; and $k$ the number of communities in the
network. In sparse networks with $m = O(n)$ and a constant number of
communities, this running time is almost linear in the size of the network.
Another important feature of our algorithm is that it can be used for either
non-overlapping or overlapping communities.
We test our algorithm using the LFR benchmark created by Lancichinetti,
Fortunato, and Radicchi specifically for the purpose of evaluating such
algorithms. Our algorithm can compete with the best of algorithms for both
non-overlapping and overlapping communities as found in the comprehensive study
of Lancichinetti and Fortunato.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 12:18:52 GMT"
},
{
"version": "v2",
"created": "Thu, 18 Dec 2014 14:52:46 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Dreier",
"Jan",
""
],
[
"Kuinke",
"Philipp",
""
],
[
"Przybylski",
"Rafael",
""
],
[
"Reidl",
"Felix",
""
],
[
"Rossmanith",
"Peter",
""
],
[
"Sikdar",
"Somnath",
""
]
] | TITLE: Overlapping Communities in Social Networks
ABSTRACT: Complex networks can be typically broken down into groups or modules.
Discovering this "community structure" is an important step in studying the
large-scale structure of networks. Many algorithms have been proposed for
community detection and benchmarks have been created to evaluate their
performance. Typically algorithms for community detection either partition the
graph (non-overlapping communities) or find node covers (overlapping
communities).
In this paper, we propose a particularly simple semi-supervised learning
algorithm for finding out communities. In essence, given the community
information of a small number of "seed nodes", the method uses random walks
from the seed nodes to uncover the community information of the whole network.
The algorithm runs in time $O(k \cdot m \cdot \log n)$, where $m$ is the number
of edges; $n$ the number of links; and $k$ the number of communities in the
network. In sparse networks with $m = O(n)$ and a constant number of
communities, this running time is almost linear in the size of the network.
Another important feature of our algorithm is that it can be used for either
non-overlapping or overlapping communities.
We test our algorithm using the LFR benchmark created by Lancichinetti,
Fortunato, and Radicchi specifically for the purpose of evaluating such
algorithms. Our algorithm can compete with the best of algorithms for both
non-overlapping and overlapping communities as found in the comprehensive study
of Lancichinetti and Fortunato.
|
1412.5617 | Shuang Song | Shuang Song, Kamalika Chaudhuri, Anand D. Sarwate | Learning from Data with Heterogeneous Noise using SGD | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider learning from data of variable quality that may be obtained from
different heterogeneous sources. Addressing learning from heterogeneous data in
its full generality is a challenging problem. In this paper, we adopt instead a
model in which data is observed through heterogeneous noise, where the noise
level reflects the quality of the data source. We study how to use stochastic
gradient algorithms to learn in this model. Our study is motivated by two
concrete examples where this problem arises naturally: learning with local
differential privacy based on data from multiple sources with different privacy
requirements, and learning from data with labels of variable quality.
The main contribution of this paper is to identify how heterogeneous noise
impacts performance. We show that given two datasets with heterogeneous noise,
the order in which to use them in standard SGD depends on the learning rate. We
propose a method for changing the learning rate as a function of the
heterogeneity, and prove new regret bounds for our method in two cases of
interest. Experiments on real data show that our method performs better than
using a single learning rate and using only the less noisy of the two datasets
when the noise level is low to moderate.
| [
{
"version": "v1",
"created": "Wed, 17 Dec 2014 21:15:06 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Song",
"Shuang",
""
],
[
"Chaudhuri",
"Kamalika",
""
],
[
"Sarwate",
"Anand D.",
""
]
] | TITLE: Learning from Data with Heterogeneous Noise using SGD
ABSTRACT: We consider learning from data of variable quality that may be obtained from
different heterogeneous sources. Addressing learning from heterogeneous data in
its full generality is a challenging problem. In this paper, we adopt instead a
model in which data is observed through heterogeneous noise, where the noise
level reflects the quality of the data source. We study how to use stochastic
gradient algorithms to learn in this model. Our study is motivated by two
concrete examples where this problem arises naturally: learning with local
differential privacy based on data from multiple sources with different privacy
requirements, and learning from data with labels of variable quality.
The main contribution of this paper is to identify how heterogeneous noise
impacts performance. We show that given two datasets with heterogeneous noise,
the order in which to use them in standard SGD depends on the learning rate. We
propose a method for changing the learning rate as a function of the
heterogeneity, and prove new regret bounds for our method in two cases of
interest. Experiments on real data show that our method performs better than
using a single learning rate and using only the less noisy of the two datasets
when the noise level is low to moderate.
|
1412.5627 | Fabricio Martins Lopes | Bruno Mendes Moro Conque and Andr\'e Yoshiaki Kashiwabara and
Fabr\'icio Martins Lopes | Feature extraction from complex networks: A case of study in genomic
sequences classification | 8 pages | null | null | null | cs.CE cs.LG q-bio.QM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work presents a new approach for classification of genomic sequences
from measurements of complex networks and information theory. For this, it is
considered the nucleotides, dinucleotides and trinucleotides of a genomic
sequence. For each of them, the entropy, sum entropy and maximum entropy values
are calculated.For each of them is also generated a network, in which the nodes
are the nucleotides, dinucleotides or trinucleotides and its edges are
estimated by observing the respective adjacency among them in the genomic
sequence. In this way, it is generated three networks, for which measures of
complex networks are extracted.These measures together with measures of
information theory comprise a feature vector representing a genomic sequence.
Thus, the feature vector is used for classification by methods such as SVM,
MultiLayer Perceptron, J48, IBK, Naive Bayes and Random Forest in order to
evaluate the proposed approach.It was adopted coding sequences, intergenic
sequences and TSS (Transcriptional Starter Sites) as datasets, for which the
better results were obtained by the Random Forest with 91.2%, followed by J48
with 89.1% and SVM with 84.8% of accuracy. These results indicate that the new
approach of feature extraction has its value, reaching good levels of
classification even considering only the genomic sequences, i.e., no other a
priori knowledge about them is considered.
| [
{
"version": "v1",
"created": "Wed, 17 Dec 2014 21:31:51 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Conque",
"Bruno Mendes Moro",
""
],
[
"Kashiwabara",
"André Yoshiaki",
""
],
[
"Lopes",
"Fabrício Martins",
""
]
] | TITLE: Feature extraction from complex networks: A case of study in genomic
sequences classification
ABSTRACT: This work presents a new approach for classification of genomic sequences
from measurements of complex networks and information theory. For this, it is
considered the nucleotides, dinucleotides and trinucleotides of a genomic
sequence. For each of them, the entropy, sum entropy and maximum entropy values
are calculated.For each of them is also generated a network, in which the nodes
are the nucleotides, dinucleotides or trinucleotides and its edges are
estimated by observing the respective adjacency among them in the genomic
sequence. In this way, it is generated three networks, for which measures of
complex networks are extracted.These measures together with measures of
information theory comprise a feature vector representing a genomic sequence.
Thus, the feature vector is used for classification by methods such as SVM,
MultiLayer Perceptron, J48, IBK, Naive Bayes and Random Forest in order to
evaluate the proposed approach.It was adopted coding sequences, intergenic
sequences and TSS (Transcriptional Starter Sites) as datasets, for which the
better results were obtained by the Random Forest with 91.2%, followed by J48
with 89.1% and SVM with 84.8% of accuracy. These results indicate that the new
approach of feature extraction has its value, reaching good levels of
classification even considering only the genomic sequences, i.e., no other a
priori knowledge about them is considered.
|
1412.5710 | Jiaqi Zhao | Jiaqi Zhao, Vitor Basto Fernandes, Licheng Jiao, Iryna Yevseyeva, Asep
Maulana, Rui Li, Thomas B\"ack, and Michael T. M. Emmerich | Multiobjective Optimization of Classifiers by Means of 3-D Convex Hull
Based Evolutionary Algorithm | 32 pages, 26 figures | null | null | null | cs.NE cs.LG | http://creativecommons.org/licenses/by/3.0/ | Finding a good classifier is a multiobjective optimization problem with
different error rates and the costs to be minimized. The receiver operating
characteristic is widely used in the machine learning community to analyze the
performance of parametric classifiers or sets of Pareto optimal classifiers. In
order to directly compare two sets of classifiers the area (or volume) under
the convex hull can be used as a scalar indicator for the performance of a set
of classifiers in receiver operating characteristic space.
Recently, the convex hull based multiobjective genetic programming algorithm
was proposed and successfully applied to maximize the convex hull area for
binary classification problems. The contribution of this paper is to extend
this algorithm for dealing with higher dimensional problem formulations. In
particular, we discuss problems where parsimony (or classifier complexity) is
stated as a third objective and multi-class classification with three different
true classification rates to be maximized.
The design of the algorithm proposed in this paper is inspired by
indicator-based evolutionary algorithms, where first a performance indicator
for a solution set is established and then a selection operator is designed
that complies with the performance indicator. In this case, the performance
indicator will be the volume under the convex hull. The algorithm is tested and
analyzed in a proof of concept study on different benchmarks that are designed
for measuring its capability to capture relevant parts of a convex hull.
Further benchmark and application studies on email classification and feature
selection round up the analysis and assess robustness and usefulness of the new
algorithm in real world settings.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 03:01:10 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Zhao",
"Jiaqi",
""
],
[
"Fernandes",
"Vitor Basto",
""
],
[
"Jiao",
"Licheng",
""
],
[
"Yevseyeva",
"Iryna",
""
],
[
"Maulana",
"Asep",
""
],
[
"Li",
"Rui",
""
],
[
"Bäck",
"Thomas",
""
],
[
"Emmerich",
"Michael T. M.",
""
]
] | TITLE: Multiobjective Optimization of Classifiers by Means of 3-D Convex Hull
Based Evolutionary Algorithm
ABSTRACT: Finding a good classifier is a multiobjective optimization problem with
different error rates and the costs to be minimized. The receiver operating
characteristic is widely used in the machine learning community to analyze the
performance of parametric classifiers or sets of Pareto optimal classifiers. In
order to directly compare two sets of classifiers the area (or volume) under
the convex hull can be used as a scalar indicator for the performance of a set
of classifiers in receiver operating characteristic space.
Recently, the convex hull based multiobjective genetic programming algorithm
was proposed and successfully applied to maximize the convex hull area for
binary classification problems. The contribution of this paper is to extend
this algorithm for dealing with higher dimensional problem formulations. In
particular, we discuss problems where parsimony (or classifier complexity) is
stated as a third objective and multi-class classification with three different
true classification rates to be maximized.
The design of the algorithm proposed in this paper is inspired by
indicator-based evolutionary algorithms, where first a performance indicator
for a solution set is established and then a selection operator is designed
that complies with the performance indicator. In this case, the performance
indicator will be the volume under the convex hull. The algorithm is tested and
analyzed in a proof of concept study on different benchmarks that are designed
for measuring its capability to capture relevant parts of a convex hull.
Further benchmark and application studies on email classification and feature
selection round up the analysis and assess robustness and usefulness of the new
algorithm in real world settings.
|
1412.5720 | David Budden | Madison Flannery, David M Budden and Alexandre Mendes | FlexDM: Enabling robust and reliable parallel data mining using WEKA | 4 pages, 2 figures | null | null | null | cs.MS cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Performing massive data mining experiments with multiple datasets and methods
is a common task faced by most bioinformatics and computational biology
laboratories. WEKA is a machine learning package designed to facilitate this
task by providing tools that allow researchers to select from several
classification methods and specific test strategies. Despite its popularity,
the current WEKA environment for batch experiments, namely Experimenter, has
four limitations that impact its usability: the selection of value ranges for
methods options lacks flexibility and is not intuitive; there is no support for
parallelisation when running large-scale data mining tasks; the XML schema is
difficult to read, necessitating the use of the Experimenter's graphical user
interface for generation and modification; and robustness is limited by the
fact that results are not saved until the last test has concluded.
FlexDM implements an interface to WEKA to run batch processing tasks in a
simple and intuitive way. In a short and easy-to-understand XML file, one can
define hundreds of tests to be performed on several datasets. FlexDM also
allows those tests to be executed asynchronously in parallel to take advantage
of multi-core processors, significantly increasing usability and productivity.
Results are saved incrementally for better robustness and reliability.
FlexDM is implemented in Java and runs on Windows, Linux and OSX. As we
encourage other researchers to explore and adopt our software, FlexDM is made
available as a pre-configured bootable reference environment. All code,
supporting documentation and usage examples are also available for download at
http://sourceforge.net/projects/flexdm.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 05:07:44 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Flannery",
"Madison",
""
],
[
"Budden",
"David M",
""
],
[
"Mendes",
"Alexandre",
""
]
] | TITLE: FlexDM: Enabling robust and reliable parallel data mining using WEKA
ABSTRACT: Performing massive data mining experiments with multiple datasets and methods
is a common task faced by most bioinformatics and computational biology
laboratories. WEKA is a machine learning package designed to facilitate this
task by providing tools that allow researchers to select from several
classification methods and specific test strategies. Despite its popularity,
the current WEKA environment for batch experiments, namely Experimenter, has
four limitations that impact its usability: the selection of value ranges for
methods options lacks flexibility and is not intuitive; there is no support for
parallelisation when running large-scale data mining tasks; the XML schema is
difficult to read, necessitating the use of the Experimenter's graphical user
interface for generation and modification; and robustness is limited by the
fact that results are not saved until the last test has concluded.
FlexDM implements an interface to WEKA to run batch processing tasks in a
simple and intuitive way. In a short and easy-to-understand XML file, one can
define hundreds of tests to be performed on several datasets. FlexDM also
allows those tests to be executed asynchronously in parallel to take advantage
of multi-core processors, significantly increasing usability and productivity.
Results are saved incrementally for better robustness and reliability.
FlexDM is implemented in Java and runs on Windows, Linux and OSX. As we
encourage other researchers to explore and adopt our software, FlexDM is made
available as a pre-configured bootable reference environment. All code,
supporting documentation and usage examples are also available for download at
http://sourceforge.net/projects/flexdm.
|
1412.5949 | Pengtao Xie | Pengtao Xie and Eric Xing | Large Scale Distributed Distance Metric Learning | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In large scale machine learning and data mining problems with high feature
dimensionality, the Euclidean distance between data points can be
uninformative, and Distance Metric Learning (DML) is often desired to learn a
proper similarity measure (using side information such as example data pairs
being similar or dissimilar). However, high dimensionality and large volume of
pairwise constraints in modern big data can lead to prohibitive computational
cost for both the original DML formulation in Xing et al. (2002) and later
extensions. In this paper, we present a distributed algorithm for DML, and a
large-scale implementation on a parameter server architecture. Our approach
builds on a parallelizable reformulation of Xing et al. (2002), and an
asynchronous stochastic gradient descent optimization procedure. To our
knowledge, this is the first distributed solution to DML, and we show that, on
a system with 256 CPU cores, our program is able to complete a DML task on a
dataset with 1 million data points, 22-thousand features, and 200 million
labeled data pairs, in 15 hours; and the learned metric shows great
effectiveness in properly measuring distances.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 17:14:34 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Xie",
"Pengtao",
""
],
[
"Xing",
"Eric",
""
]
] | TITLE: Large Scale Distributed Distance Metric Learning
ABSTRACT: In large scale machine learning and data mining problems with high feature
dimensionality, the Euclidean distance between data points can be
uninformative, and Distance Metric Learning (DML) is often desired to learn a
proper similarity measure (using side information such as example data pairs
being similar or dissimilar). However, high dimensionality and large volume of
pairwise constraints in modern big data can lead to prohibitive computational
cost for both the original DML formulation in Xing et al. (2002) and later
extensions. In this paper, we present a distributed algorithm for DML, and a
large-scale implementation on a parameter server architecture. Our approach
builds on a parallelizable reformulation of Xing et al. (2002), and an
asynchronous stochastic gradient descent optimization procedure. To our
knowledge, this is the first distributed solution to DML, and we show that, on
a system with 256 CPU cores, our program is able to complete a DML task on a
dataset with 1 million data points, 22-thousand features, and 200 million
labeled data pairs, in 15 hours; and the learned metric shows great
effectiveness in properly measuring distances.
|
1412.5968 | Andrew Lan | Andrew S. Lan, Christoph Studer, Richard G. Baraniuk | Quantized Matrix Completion for Personalized Learning | null | In Proc. 7th Intl. Conf. on Educational Data Mining, pages
280-283, July 2014 | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recently proposed SPARse Factor Analysis (SPARFA) framework for
personalized learning performs factor analysis on ordinal or binary-valued
(e.g., correct/incorrect) graded learner responses to questions. The underlying
factors are termed "concepts" (or knowledge components) and are used for
learning analytics (LA), the estimation of learner concept-knowledge profiles,
and for content analytics (CA), the estimation of question-concept associations
and question difficulties. While SPARFA is a powerful tool for LA and CA, it
requires a number of algorithm parameters (including the number of concepts),
which are difficult to determine in practice. In this paper, we propose
SPARFA-Lite, a convex optimization-based method for LA that builds on matrix
completion, which only requires a single algorithm parameter and enables us to
automatically identify the required number of concepts. Using a variety of
educational datasets, we demonstrate that SPARFALite (i) achieves comparable
performance in predicting unobserved learner responses to existing methods,
including item response theory (IRT) and SPARFA, and (ii) is computationally
more efficient.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 17:48:17 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Lan",
"Andrew S.",
""
],
[
"Studer",
"Christoph",
""
],
[
"Baraniuk",
"Richard G.",
""
]
] | TITLE: Quantized Matrix Completion for Personalized Learning
ABSTRACT: The recently proposed SPARse Factor Analysis (SPARFA) framework for
personalized learning performs factor analysis on ordinal or binary-valued
(e.g., correct/incorrect) graded learner responses to questions. The underlying
factors are termed "concepts" (or knowledge components) and are used for
learning analytics (LA), the estimation of learner concept-knowledge profiles,
and for content analytics (CA), the estimation of question-concept associations
and question difficulties. While SPARFA is a powerful tool for LA and CA, it
requires a number of algorithm parameters (including the number of concepts),
which are difficult to determine in practice. In this paper, we propose
SPARFA-Lite, a convex optimization-based method for LA that builds on matrix
completion, which only requires a single algorithm parameter and enables us to
automatically identify the required number of concepts. Using a variety of
educational datasets, we demonstrate that SPARFALite (i) achieves comparable
performance in predicting unobserved learner responses to existing methods,
including item response theory (IRT) and SPARFA, and (ii) is computationally
more efficient.
|
1412.5984 | Swakkhar Shatabda | Muktadir Hossain, Tajkia Tasnim, Swakkhar Shatabda and Dewan M. Farid | Stochastic Local Search for Pattern Set Mining | The 8th International Conference on Software, Knowledge, Information
Management and Applications (SKIMA 2014) | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Local search methods can quickly find good quality solutions in cases where
systematic search methods might take a large amount of time. Moreover, in the
context of pattern set mining, exhaustive search methods are not applicable due
to the large search space they have to explore. In this paper, we propose the
application of stochastic local search to solve the pattern set mining.
Specifically, to the task of concept learning. We applied a number of local
search algorithms on a standard benchmark instances for pattern set mining and
the results show the potentials for further exploration.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 18:16:52 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Hossain",
"Muktadir",
""
],
[
"Tasnim",
"Tajkia",
""
],
[
"Shatabda",
"Swakkhar",
""
],
[
"Farid",
"Dewan M.",
""
]
] | TITLE: Stochastic Local Search for Pattern Set Mining
ABSTRACT: Local search methods can quickly find good quality solutions in cases where
systematic search methods might take a large amount of time. Moreover, in the
context of pattern set mining, exhaustive search methods are not applicable due
to the large search space they have to explore. In this paper, we propose the
application of stochastic local search to solve the pattern set mining.
Specifically, to the task of concept learning. We applied a number of local
search algorithms on a standard benchmark instances for pattern set mining and
the results show the potentials for further exploration.
|
1412.6018 | Rachada Kongkachandra Asst.Prof.Dr. | Sirisak Visessenee, Sanparith Marukatat, and Rachada Kongkachandra | Automatic Training Data Synthesis for Handwriting Recognition Using the
Structural Crossing-Over Technique | 8 pages, 6 figures | International Journal of Artificial Intelligence & Applications
(IJAIA), Vol. 5, No. 5, September 2014 | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The paper presents a novel technique called "Structural Crossing-Over" to
synthesize qualified data for training machine learning-based handwriting
recognition. The proposed technique can provide a greater variety of patterns
of training data than the existing approaches such as elastic distortion and
tangent-based affine transformation. A couple of training characters are
chosen, then they are analyzed by their similar and different structures, and
finally are crossed over to generate the new characters. The experiments are
set to compare the performances of tangent-based affine transformation and the
proposed approach in terms of the variety of generated characters and percent
of recognition errors. The standard MNIST corpus including 60,000 training
characters and 10,000 test characters is employed in the experiments. The
proposed technique uses 1,000 characters to synthesize 60,000 characters, and
then uses these data to train and test the benchmark handwriting recognition
system that exploits Histogram of Gradient (HOG) as features and Support Vector
Machine (SVM) as recognizer. The experimental result yields 8.06% of errors. It
significantly outperforms the tangent-based affine transformation and the
original MNIST training data, which are 11.74% and 16.55%, respectively.
| [
{
"version": "v1",
"created": "Thu, 9 Oct 2014 04:32:20 GMT"
}
] | 2014-12-19T00:00:00 | [
[
"Visessenee",
"Sirisak",
""
],
[
"Marukatat",
"Sanparith",
""
],
[
"Kongkachandra",
"Rachada",
""
]
] | TITLE: Automatic Training Data Synthesis for Handwriting Recognition Using the
Structural Crossing-Over Technique
ABSTRACT: The paper presents a novel technique called "Structural Crossing-Over" to
synthesize qualified data for training machine learning-based handwriting
recognition. The proposed technique can provide a greater variety of patterns
of training data than the existing approaches such as elastic distortion and
tangent-based affine transformation. A couple of training characters are
chosen, then they are analyzed by their similar and different structures, and
finally are crossed over to generate the new characters. The experiments are
set to compare the performances of tangent-based affine transformation and the
proposed approach in terms of the variety of generated characters and percent
of recognition errors. The standard MNIST corpus including 60,000 training
characters and 10,000 test characters is employed in the experiments. The
proposed technique uses 1,000 characters to synthesize 60,000 characters, and
then uses these data to train and test the benchmark handwriting recognition
system that exploits Histogram of Gradient (HOG) as features and Support Vector
Machine (SVM) as recognizer. The experimental result yields 8.06% of errors. It
significantly outperforms the tangent-based affine transformation and the
original MNIST training data, which are 11.74% and 16.55%, respectively.
|
1302.0296 | Navid Naderializadeh | Navid Naderializadeh, A. Salman Avestimehr | Interference Networks with No CSIT: Impact of Topology | Accepted for Publication in IEEE Transactions on Information Theory | null | null | null | cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider partially-connected $K$-user interference networks, where the
transmitters have no knowledge about the channel gain values, but they are
aware of network topology (or connectivity). We introduce several linear
algebraic and graph theoretic concepts to derive new topology-based outer
bounds and inner bounds on the symmetric degrees-of-freedom (DoF) of these
networks. We evaluate our bounds for two classes of networks to demonstrate
their tightness for most networks in these classes, quantify the gain of our
inner bounds over benchmark interference management strategies, and illustrate
the effect of network topology on these gains.
| [
{
"version": "v1",
"created": "Fri, 1 Feb 2013 21:22:16 GMT"
},
{
"version": "v2",
"created": "Wed, 17 Dec 2014 07:45:20 GMT"
}
] | 2014-12-18T00:00:00 | [
[
"Naderializadeh",
"Navid",
""
],
[
"Avestimehr",
"A. Salman",
""
]
] | TITLE: Interference Networks with No CSIT: Impact of Topology
ABSTRACT: We consider partially-connected $K$-user interference networks, where the
transmitters have no knowledge about the channel gain values, but they are
aware of network topology (or connectivity). We introduce several linear
algebraic and graph theoretic concepts to derive new topology-based outer
bounds and inner bounds on the symmetric degrees-of-freedom (DoF) of these
networks. We evaluate our bounds for two classes of networks to demonstrate
their tightness for most networks in these classes, quantify the gain of our
inner bounds over benchmark interference management strategies, and illustrate
the effect of network topology on these gains.
|
1309.4796 | Lijun Peng | Lijun Peng, Luis Carvalho | Bayesian Degree-Corrected Stochastic Blockmodels for Community Detection | 23 pages, 9 figures | null | null | null | stat.ME cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Community detection in networks has drawn much attention in diverse fields,
especially social sciences. Given its significance, there has been a large body
of literature with approaches from many fields. Here we present a statistical
framework that is representative, extensible, and that yields an estimator with
good properties. Our proposed approach considers a stochastic blockmodel based
on a logistic regression formulation with node correction terms. We follow a
Bayesian approach that explicitly captures the community behavior via prior
specification. We further adopt a data augmentation strategy with latent
Polya-Gamma variables to obtain posterior samples. We conduct inference based
on a principled, canonically mapped centroid estimator that formally addresses
label non-identifiability and captures representative community assignments. We
demonstrate the proposed model and estimation on real-world as well as
simulated benchmark networks and show that the proposed model and estimator are
more flexible, representative, and yield smaller error rates when compared to
the MAP estimator from classical degree-corrected stochastic blockmodels.
| [
{
"version": "v1",
"created": "Wed, 18 Sep 2013 21:14:45 GMT"
},
{
"version": "v2",
"created": "Wed, 17 Dec 2014 17:50:02 GMT"
}
] | 2014-12-18T00:00:00 | [
[
"Peng",
"Lijun",
""
],
[
"Carvalho",
"Luis",
""
]
] | TITLE: Bayesian Degree-Corrected Stochastic Blockmodels for Community Detection
ABSTRACT: Community detection in networks has drawn much attention in diverse fields,
especially social sciences. Given its significance, there has been a large body
of literature with approaches from many fields. Here we present a statistical
framework that is representative, extensible, and that yields an estimator with
good properties. Our proposed approach considers a stochastic blockmodel based
on a logistic regression formulation with node correction terms. We follow a
Bayesian approach that explicitly captures the community behavior via prior
specification. We further adopt a data augmentation strategy with latent
Polya-Gamma variables to obtain posterior samples. We conduct inference based
on a principled, canonically mapped centroid estimator that formally addresses
label non-identifiability and captures representative community assignments. We
demonstrate the proposed model and estimation on real-world as well as
simulated benchmark networks and show that the proposed model and estimator are
more flexible, representative, and yield smaller error rates when compared to
the MAP estimator from classical degree-corrected stochastic blockmodels.
|
1412.5448 | Micka\"el Poussevin | Micka\"el Poussevin and Vincent Guigue and Patrick Gallinari | Extended Recommendation Framework: Generating the Text of a User Review
as a Personalized Summary | null | null | null | null | cs.IR cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose to augment rating based recommender systems by providing the user
with additional information which might help him in his choice or in the
understanding of the recommendation. We consider here as a new task, the
generation of personalized reviews associated to items. We use an extractive
summary formulation for generating these reviews. We also show that the two
information sources, ratings and items could be used both for estimating
ratings and for generating summaries, leading to improved performance for each
system compared to the use of a single source. Besides these two contributions,
we show how a personalized polarity classifier can integrate the rating and
textual aspects. Overall, the proposed system offers the user three
personalized hints for a recommendation: rating, text and polarity. We evaluate
these three components on two datasets using appropriate measures for each
task.
| [
{
"version": "v1",
"created": "Wed, 17 Dec 2014 15:46:28 GMT"
}
] | 2014-12-18T00:00:00 | [
[
"Poussevin",
"Mickaël",
""
],
[
"Guigue",
"Vincent",
""
],
[
"Gallinari",
"Patrick",
""
]
] | TITLE: Extended Recommendation Framework: Generating the Text of a User Review
as a Personalized Summary
ABSTRACT: We propose to augment rating based recommender systems by providing the user
with additional information which might help him in his choice or in the
understanding of the recommendation. We consider here as a new task, the
generation of personalized reviews associated to items. We use an extractive
summary formulation for generating these reviews. We also show that the two
information sources, ratings and items could be used both for estimating
ratings and for generating summaries, leading to improved performance for each
system compared to the use of a single source. Besides these two contributions,
we show how a personalized polarity classifier can integrate the rating and
textual aspects. Overall, the proposed system offers the user three
personalized hints for a recommendation: rating, text and polarity. We evaluate
these three components on two datasets using appropriate measures for each
task.
|
1412.5488 | Ashirbani Saha | Ashirbani Saha, Q. M. Jonathan Wu | Full-reference image quality assessment by combining global and local
distortion measures | 31 pages, 8 figures, 2 tables | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Full-reference image quality assessment (FR-IQA) techniques compare a
reference and a distorted/test image and predict the perceptual quality of the
test image in terms of a scalar value representing an objective score. The
evaluation of FR-IQA techniques is carried out by comparing the objective
scores from the techniques with the subjective scores (obtained from human
observers) provided in the image databases used for the IQA. Hence, we
reasonably assume that the goal of a human observer is to rate the distortion
present in the test image. The goal oriented tasks are processed by the human
visual system (HVS) through top-down processing which actively searches for
local distortions driven by the goal. Therefore local distortion measures in an
image are important for the top-down processing. At the same time, bottom-up
processing also takes place signifying spontaneous visual functions in the HVS.
To account for this, global perceptual features can be used. Therefore, we
hypothesize that the resulting objective score for an image can be derived from
the combination of local and global distortion measures calculated from the
reference and test images. We calculate the local distortion by measuring the
local correlation differences from the gradient and contrast information. For
global distortion, dissimilarity of the saliency maps computed from a bottom-up
model of saliency is used. The motivation behind the proposed approach has been
thoroughly discussed, accompanied by an intuitive analysis. Finally,
experiments are conducted in six benchmark databases suggesting the
effectiveness of the proposed approach that achieves competitive performance
with the state-of-the-art methods providing an improvement in the overall
performance.
| [
{
"version": "v1",
"created": "Wed, 17 Dec 2014 17:15:12 GMT"
}
] | 2014-12-18T00:00:00 | [
[
"Saha",
"Ashirbani",
""
],
[
"Wu",
"Q. M. Jonathan",
""
]
] | TITLE: Full-reference image quality assessment by combining global and local
distortion measures
ABSTRACT: Full-reference image quality assessment (FR-IQA) techniques compare a
reference and a distorted/test image and predict the perceptual quality of the
test image in terms of a scalar value representing an objective score. The
evaluation of FR-IQA techniques is carried out by comparing the objective
scores from the techniques with the subjective scores (obtained from human
observers) provided in the image databases used for the IQA. Hence, we
reasonably assume that the goal of a human observer is to rate the distortion
present in the test image. The goal oriented tasks are processed by the human
visual system (HVS) through top-down processing which actively searches for
local distortions driven by the goal. Therefore local distortion measures in an
image are important for the top-down processing. At the same time, bottom-up
processing also takes place signifying spontaneous visual functions in the HVS.
To account for this, global perceptual features can be used. Therefore, we
hypothesize that the resulting objective score for an image can be derived from
the combination of local and global distortion measures calculated from the
reference and test images. We calculate the local distortion by measuring the
local correlation differences from the gradient and contrast information. For
global distortion, dissimilarity of the saliency maps computed from a bottom-up
model of saliency is used. The motivation behind the proposed approach has been
thoroughly discussed, accompanied by an intuitive analysis. Finally,
experiments are conducted in six benchmark databases suggesting the
effectiveness of the proposed approach that achieves competitive performance
with the state-of-the-art methods providing an improvement in the overall
performance.
|
1412.5513 | Engelbert Mephu Nguifo | Cyrine Arouri, Engelbert Mephu Nguifo, Sabeur Aridhi, C\'ecile
Roucelle, Gaelle Bonnet-Loosli, Norbert Tsopz\'e | Towards a constructive multilayer perceptron for regression task using
non-parametric clustering. A case study of Photo-Z redshift reconstruction | null | null | null | null | cs.NE cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The choice of architecture of artificial neuron network (ANN) is still a
challenging task that users face every time. It greatly affects the accuracy of
the built network. In fact there is no optimal method that is applicable to
various implementations at the same time. In this paper we propose a method to
construct ANN based on clustering, that resolves the problems of random and ad
hoc approaches for multilayer ANN architecture. Our method can be applied to
regression problems. Experimental results obtained with different datasets,
reveals the efficiency of our method.
| [
{
"version": "v1",
"created": "Wed, 17 Dec 2014 18:36:23 GMT"
}
] | 2014-12-18T00:00:00 | [
[
"Arouri",
"Cyrine",
""
],
[
"Nguifo",
"Engelbert Mephu",
""
],
[
"Aridhi",
"Sabeur",
""
],
[
"Roucelle",
"Cécile",
""
],
[
"Bonnet-Loosli",
"Gaelle",
""
],
[
"Tsopzé",
"Norbert",
""
]
] | TITLE: Towards a constructive multilayer perceptron for regression task using
non-parametric clustering. A case study of Photo-Z redshift reconstruction
ABSTRACT: The choice of architecture of artificial neuron network (ANN) is still a
challenging task that users face every time. It greatly affects the accuracy of
the built network. In fact there is no optimal method that is applicable to
various implementations at the same time. In this paper we propose a method to
construct ANN based on clustering, that resolves the problems of random and ad
hoc approaches for multilayer ANN architecture. Our method can be applied to
regression problems. Experimental results obtained with different datasets,
reveals the efficiency of our method.
|
1411.4044 | Todor Ivanov | Todor Ivanov, Raik Niemann, Sead Izberovic, Marten Rosselli, Karsten
Tolle, Roberto V. Zicari | Benchmarking DataStax Enterprise/Cassandra with HiBench | arXiv admin note: substantial text overlap with arXiv:1411.3811 | null | null | Technical Report No. 2014-2 | cs.DC cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This report evaluates the new analytical capabilities of DataStax Enterprise
(DSE) [1] through the use of standard Hadoop workloads. In particular, we run
experiments with CPU and I/O bound micro-benchmarks as well as OLAP-style
analytical query workloads. The performed tests should show that DSE is capable
of successfully executing Hadoop applications without the need to adapt them
for the underlying Cassandra distributed storage system [2]. Due to the
Cassandra File System (CFS) [3], which supports the Hadoop Distributed File
System API, Hadoop stack applications should seamlessly run in DSE. The report
is structured as follows: Section 2 provides a brief description of the
technologies involved in our study. An overview of our used hardware and
software components of the experimental environment is given in Section 3. Our
benchmark methodology is defined in Section 4. The performed experiments
together with the evaluation of the results are presented in Section 5.
Finally, Section 6 concludes with lessons learned.
| [
{
"version": "v1",
"created": "Fri, 14 Nov 2014 07:17:54 GMT"
},
{
"version": "v2",
"created": "Tue, 16 Dec 2014 14:34:47 GMT"
}
] | 2014-12-17T00:00:00 | [
[
"Ivanov",
"Todor",
""
],
[
"Niemann",
"Raik",
""
],
[
"Izberovic",
"Sead",
""
],
[
"Rosselli",
"Marten",
""
],
[
"Tolle",
"Karsten",
""
],
[
"Zicari",
"Roberto V.",
""
]
] | TITLE: Benchmarking DataStax Enterprise/Cassandra with HiBench
ABSTRACT: This report evaluates the new analytical capabilities of DataStax Enterprise
(DSE) [1] through the use of standard Hadoop workloads. In particular, we run
experiments with CPU and I/O bound micro-benchmarks as well as OLAP-style
analytical query workloads. The performed tests should show that DSE is capable
of successfully executing Hadoop applications without the need to adapt them
for the underlying Cassandra distributed storage system [2]. Due to the
Cassandra File System (CFS) [3], which supports the Hadoop Distributed File
System API, Hadoop stack applications should seamlessly run in DSE. The report
is structured as follows: Section 2 provides a brief description of the
technologies involved in our study. An overview of our used hardware and
software components of the experimental environment is given in Section 3. Our
benchmark methodology is defined in Section 4. The performed experiments
together with the evaluation of the results are presented in Section 5.
Finally, Section 6 concludes with lessons learned.
|
1412.4842 | Mingjie Tang | Mingjie Tang, Ruby Y.Tahboub, Walid G.Are, Mikhail J. Atallah,
Qutaibah M. Malluhi, Mourad Ouzzani, and Yasin N. Silva | Similarity Group-by Operators for Multi-dimensional Relational Data | submit to TKDE | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The SQL group-by operator plays an important role in summarizing and
aggregating large datasets in a data analytic stack.While the standard group-by
operator, which is based on equality, is useful in several applications,
allowing similarity aware grouping provides a more realistic view on real-world
data that could lead to better insights. The Similarity SQL-based Group-By
operator (SGB, for short) extends the semantics of the standard SQL Group-by by
grouping data with similar but not necessarily equal values. While existing
similarity-based grouping operators efficiently materialize this approximate
semantics, they primarily focus on one-dimensional attributes and treat
multidimensional attributes independently. However, correlated attributes, such
as in spatial data, are processed independently, and hence, groups in the
multidimensional space are not detected properly. To address this problem, we
introduce two new SGB operators for multidimensional data. The first operator
is the clique (or distance-to-all) SGB, where all the tuples in a group are
within some distance from each other. The second operator is the
distance-to-any SGB, where a tuple belongs to a group if the tuple is within
some distance from any other tuple in the group. We implement and test the new
SGB operators and their algorithms inside PostgreSQL. The overhead introduced
by these operators proves to be minimal and the execution times are comparable
to those of the standard Group-by. The experimental study, based on TPC-H and a
social check-in data, demonstrates that the proposed algorithms can achieve up
to three orders of magnitude enhancement in performance over baseline methods
developed to solve the same problem.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 00:27:52 GMT"
}
] | 2014-12-17T00:00:00 | [
[
"Tang",
"Mingjie",
""
],
[
"Tahboub",
"Ruby Y.",
""
],
[
"Are",
"Walid G.",
""
],
[
"Atallah",
"Mikhail J.",
""
],
[
"Malluhi",
"Qutaibah M.",
""
],
[
"Ouzzani",
"Mourad",
""
],
[
"Silva",
"Yasin N.",
""
]
] | TITLE: Similarity Group-by Operators for Multi-dimensional Relational Data
ABSTRACT: The SQL group-by operator plays an important role in summarizing and
aggregating large datasets in a data analytic stack.While the standard group-by
operator, which is based on equality, is useful in several applications,
allowing similarity aware grouping provides a more realistic view on real-world
data that could lead to better insights. The Similarity SQL-based Group-By
operator (SGB, for short) extends the semantics of the standard SQL Group-by by
grouping data with similar but not necessarily equal values. While existing
similarity-based grouping operators efficiently materialize this approximate
semantics, they primarily focus on one-dimensional attributes and treat
multidimensional attributes independently. However, correlated attributes, such
as in spatial data, are processed independently, and hence, groups in the
multidimensional space are not detected properly. To address this problem, we
introduce two new SGB operators for multidimensional data. The first operator
is the clique (or distance-to-all) SGB, where all the tuples in a group are
within some distance from each other. The second operator is the
distance-to-any SGB, where a tuple belongs to a group if the tuple is within
some distance from any other tuple in the group. We implement and test the new
SGB operators and their algorithms inside PostgreSQL. The overhead introduced
by these operators proves to be minimal and the execution times are comparable
to those of the standard Group-by. The experimental study, based on TPC-H and a
social check-in data, demonstrates that the proposed algorithms can achieve up
to three orders of magnitude enhancement in performance over baseline methods
developed to solve the same problem.
|
1412.4864 | Philip Bachman | Philip Bachman and Ouais Alsharif and Doina Precup | Learning with Pseudo-Ensembles | To appear in Advances in Neural Information Processing Systems 27
(NIPS 2014), Advances in Neural Information Processing Systems 27, Dec. 2014 | null | null | null | stat.ML cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We formalize the notion of a pseudo-ensemble, a (possibly infinite)
collection of child models spawned from a parent model by perturbing it
according to some noise process. E.g., dropout (Hinton et. al, 2012) in a deep
neural network trains a pseudo-ensemble of child subnetworks generated by
randomly masking nodes in the parent network. We present a novel regularizer
based on making the behavior of a pseudo-ensemble robust with respect to the
noise process generating it. In the fully-supervised setting, our regularizer
matches the performance of dropout. But, unlike dropout, our regularizer
naturally extends to the semi-supervised setting, where it produces
state-of-the-art results. We provide a case study in which we transform the
Recursive Neural Tensor Network of (Socher et. al, 2013) into a
pseudo-ensemble, which significantly improves its performance on a real-world
sentiment analysis benchmark.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 02:55:05 GMT"
}
] | 2014-12-17T00:00:00 | [
[
"Bachman",
"Philip",
""
],
[
"Alsharif",
"Ouais",
""
],
[
"Precup",
"Doina",
""
]
] | TITLE: Learning with Pseudo-Ensembles
ABSTRACT: We formalize the notion of a pseudo-ensemble, a (possibly infinite)
collection of child models spawned from a parent model by perturbing it
according to some noise process. E.g., dropout (Hinton et. al, 2012) in a deep
neural network trains a pseudo-ensemble of child subnetworks generated by
randomly masking nodes in the parent network. We present a novel regularizer
based on making the behavior of a pseudo-ensemble robust with respect to the
noise process generating it. In the fully-supervised setting, our regularizer
matches the performance of dropout. But, unlike dropout, our regularizer
naturally extends to the semi-supervised setting, where it produces
state-of-the-art results. We provide a case study in which we transform the
Recursive Neural Tensor Network of (Socher et. al, 2013) into a
pseudo-ensemble, which significantly improves its performance on a real-world
sentiment analysis benchmark.
|
1412.4967 | Anthony Knittel | Anthony Knittel, Alan Blair | Sparse, guided feature connections in an Abstract Deep Network | null | null | null | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a technique for developing a network of re-used features, where
the topology is formed using a coarse learning method, that allows
gradient-descent fine tuning, known as an Abstract Deep Network (ADN). New
features are built based on observed co-occurrences, and the network is
maintained using a selection process related to evolutionary algorithms. This
allows coarse ex- ploration of the problem space, effective for irregular
domains, while gradient descent allows pre- cise solutions. Accuracy on
standard UCI and Protein-Structure Prediction problems is comparable with
benchmark SVM and optimized GBML approaches, and shows scalability for
addressing large problems. The discrete implementation is symbolic, allowing
interpretability, while the continuous method using fine-tuning shows improved
accuracy. The binary multiplexer problem is explored, as an irregular domain
that does not support gradient descent learning, showing solution to the bench-
mark 135-bit problem. A convolutional implementation is demonstrated on image
classification, showing an error-rate of 0.79% on the MNIST problem, without a
pre-defined topology. The ADN system provides a method for developing a very
sparse, deep feature topology, based on observed relationships between
features, that is able to find solutions in irregular domains, and initialize a
network prior to gradient descent learning.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 12:04:35 GMT"
}
] | 2014-12-17T00:00:00 | [
[
"Knittel",
"Anthony",
""
],
[
"Blair",
"Alan",
""
]
] | TITLE: Sparse, guided feature connections in an Abstract Deep Network
ABSTRACT: We present a technique for developing a network of re-used features, where
the topology is formed using a coarse learning method, that allows
gradient-descent fine tuning, known as an Abstract Deep Network (ADN). New
features are built based on observed co-occurrences, and the network is
maintained using a selection process related to evolutionary algorithms. This
allows coarse ex- ploration of the problem space, effective for irregular
domains, while gradient descent allows pre- cise solutions. Accuracy on
standard UCI and Protein-Structure Prediction problems is comparable with
benchmark SVM and optimized GBML approaches, and shows scalability for
addressing large problems. The discrete implementation is symbolic, allowing
interpretability, while the continuous method using fine-tuning shows improved
accuracy. The binary multiplexer problem is explored, as an irregular domain
that does not support gradient descent learning, showing solution to the bench-
mark 135-bit problem. A convolutional implementation is demonstrated on image
classification, showing an error-rate of 0.79% on the MNIST problem, without a
pre-defined topology. The ADN system provides a method for developing a very
sparse, deep feature topology, based on observed relationships between
features, that is able to find solutions in irregular domains, and initialize a
network prior to gradient descent learning.
|
1412.5067 | Anton Eremeev | A. V. Eremeev, Ju. V. Kovalenko | Analysis of Optimal Recombination in Genetic Algorithm for a Scheduling
Problem with Setups | 13 pages, in Russian | null | null | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we perform an experimental study of optimal recombination
operator for makespan minimization problem on single machine with
sequence-dependent setup times ($1|s_{vu}|C_{\max}$). The computational
experiment on benchmark problems from TSPLIB library indicates practical
applicability of optimal recombination in crossover operator of genetic
algorithm for $1|s_{vu}|C_{\max}$.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 16:29:34 GMT"
}
] | 2014-12-17T00:00:00 | [
[
"Eremeev",
"A. V.",
""
],
[
"Kovalenko",
"Ju. V.",
""
]
] | TITLE: Analysis of Optimal Recombination in Genetic Algorithm for a Scheduling
Problem with Setups
ABSTRACT: In this paper, we perform an experimental study of optimal recombination
operator for makespan minimization problem on single machine with
sequence-dependent setup times ($1|s_{vu}|C_{\max}$). The computational
experiment on benchmark problems from TSPLIB library indicates practical
applicability of optimal recombination in crossover operator of genetic
algorithm for $1|s_{vu}|C_{\max}$.
|
1412.5104 | Angjoo Kanazawa | Angjoo Kanazawa, Abhishek Sharma, David Jacobs | Locally Scale-Invariant Convolutional Neural Networks | Deep Learning and Representation Learning Workshop: NIPS 2014 | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Convolutional Neural Networks (ConvNets) have shown excellent results on many
visual classification tasks. With the exception of ImageNet, these datasets are
carefully crafted such that objects are well-aligned at similar scales.
Naturally, the feature learning problem gets more challenging as the amount of
variation in the data increases, as the models have to learn to be invariant to
certain changes in appearance. Recent results on the ImageNet dataset show that
given enough data, ConvNets can learn such invariances producing very
discriminative features [1]. But could we do more: use less parameters, less
data, learn more discriminative features, if certain invariances were built
into the learning process? In this paper we present a simple model that allows
ConvNets to learn features in a locally scale-invariant manner without
increasing the number of model parameters. We show on a modified MNIST dataset
that when faced with scale variation, building in scale-invariance allows
ConvNets to learn more discriminative features with reduced chances of
over-fitting.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 18:09:34 GMT"
}
] | 2014-12-17T00:00:00 | [
[
"Kanazawa",
"Angjoo",
""
],
[
"Sharma",
"Abhishek",
""
],
[
"Jacobs",
"David",
""
]
] | TITLE: Locally Scale-Invariant Convolutional Neural Networks
ABSTRACT: Convolutional Neural Networks (ConvNets) have shown excellent results on many
visual classification tasks. With the exception of ImageNet, these datasets are
carefully crafted such that objects are well-aligned at similar scales.
Naturally, the feature learning problem gets more challenging as the amount of
variation in the data increases, as the models have to learn to be invariant to
certain changes in appearance. Recent results on the ImageNet dataset show that
given enough data, ConvNets can learn such invariances producing very
discriminative features [1]. But could we do more: use less parameters, less
data, learn more discriminative features, if certain invariances were built
into the learning process? In this paper we present a simple model that allows
ConvNets to learn features in a locally scale-invariant manner without
increasing the number of model parameters. We show on a modified MNIST dataset
that when faced with scale variation, building in scale-invariance allows
ConvNets to learn more discriminative features with reduced chances of
over-fitting.
|
1409.3215 | Ilya Sutskever | Ilya Sutskever and Oriol Vinyals and Quoc V. Le | Sequence to Sequence Learning with Neural Networks | 9 pages | null | null | null | cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep Neural Networks (DNNs) are powerful models that have achieved excellent
performance on difficult learning tasks. Although DNNs work well whenever large
labeled training sets are available, they cannot be used to map sequences to
sequences. In this paper, we present a general end-to-end approach to sequence
learning that makes minimal assumptions on the sequence structure. Our method
uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to
a vector of a fixed dimensionality, and then another deep LSTM to decode the
target sequence from the vector. Our main result is that on an English to
French translation task from the WMT'14 dataset, the translations produced by
the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's
BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did
not have difficulty on long sentences. For comparison, a phrase-based SMT
system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM
to rerank the 1000 hypotheses produced by the aforementioned SMT system, its
BLEU score increases to 36.5, which is close to the previous best result on
this task. The LSTM also learned sensible phrase and sentence representations
that are sensitive to word order and are relatively invariant to the active and
the passive voice. Finally, we found that reversing the order of the words in
all source sentences (but not target sentences) improved the LSTM's performance
markedly, because doing so introduced many short term dependencies between the
source and the target sentence which made the optimization problem easier.
| [
{
"version": "v1",
"created": "Wed, 10 Sep 2014 19:55:35 GMT"
},
{
"version": "v2",
"created": "Wed, 29 Oct 2014 12:13:17 GMT"
},
{
"version": "v3",
"created": "Sun, 14 Dec 2014 20:59:51 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Sutskever",
"Ilya",
""
],
[
"Vinyals",
"Oriol",
""
],
[
"Le",
"Quoc V.",
""
]
] | TITLE: Sequence to Sequence Learning with Neural Networks
ABSTRACT: Deep Neural Networks (DNNs) are powerful models that have achieved excellent
performance on difficult learning tasks. Although DNNs work well whenever large
labeled training sets are available, they cannot be used to map sequences to
sequences. In this paper, we present a general end-to-end approach to sequence
learning that makes minimal assumptions on the sequence structure. Our method
uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to
a vector of a fixed dimensionality, and then another deep LSTM to decode the
target sequence from the vector. Our main result is that on an English to
French translation task from the WMT'14 dataset, the translations produced by
the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's
BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did
not have difficulty on long sentences. For comparison, a phrase-based SMT
system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM
to rerank the 1000 hypotheses produced by the aforementioned SMT system, its
BLEU score increases to 36.5, which is close to the previous best result on
this task. The LSTM also learned sensible phrase and sentence representations
that are sensitive to word order and are relatively invariant to the active and
the passive voice. Finally, we found that reversing the order of the words in
all source sentences (but not target sentences) improved the LSTM's performance
markedly, because doing so introduced many short term dependencies between the
source and the target sentence which made the optimization problem easier.
|
1412.4183 | Eugene Borovikov | Eugene Borovikov | A survey of modern optical character recognition techniques | Technical report surveying OCR/ICR and document understanding methods
as of 2004.It contains 38 pages, numerous figures, 93 references, and
provides a table of contents | null | null | null | cs.CV | http://creativecommons.org/licenses/publicdomain/ | This report explores the latest advances in the field of digital document
recognition. With the focus on printed document imagery, we discuss the major
developments in optical character recognition (OCR) and document image
enhancement/restoration in application to Latin and non-Latin scripts. In
addition, we review and discuss the available technologies for hand-written
document recognition. In this report, we also provide some company-accumulated
benchmark results on available OCR engines.
| [
{
"version": "v1",
"created": "Sat, 13 Dec 2014 02:39:09 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Borovikov",
"Eugene",
""
]
] | TITLE: A survey of modern optical character recognition techniques
ABSTRACT: This report explores the latest advances in the field of digital document
recognition. With the focus on printed document imagery, we discuss the major
developments in optical character recognition (OCR) and document image
enhancement/restoration in application to Latin and non-Latin scripts. In
addition, we review and discuss the available technologies for hand-written
document recognition. In this report, we also provide some company-accumulated
benchmark results on available OCR engines.
|
1412.4196 | Yuanting Hu | Yuan-Ting Hu and Yen-Yu Lin and Hsin-Yi Chen and Kuang-Jui Hsu and
Bing-Yu Chen | Descriptor Ensemble: An Unsupervised Approach to Descriptor Fusion in
the Homography Space | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the aim to improve the performance of feature matching, we present an
unsupervised approach to fuse various local descriptors in the space of
homographies. Inspired by the observation that the homographies of correct
feature correspondences vary smoothly along the spatial domain, our approach
stands on the unsupervised nature of feature matching, and can select a good
descriptor for matching each feature point. Specifically, the homography space
serves as the common domain, in which a correspondence obtained by any
descriptor is considered as a point, for integrating various heterogeneous
descriptors. Both geometric coherence and spatial continuity among
correspondences are considered via computing their geodesic distances in the
space. In this way, mutual verification across different descriptors is
allowed, and correct correspondences will be highlighted with a high degree of
consistency (i.e., short geodesic distances here). It follows that one-class
SVM can be applied to identifying these correct correspondences, and boosts the
performance of feature matching. The proposed approach is comprehensively
compared with the state-of-the-art approaches, and evaluated on four benchmarks
of image matching. The promising results manifest its effectiveness.
| [
{
"version": "v1",
"created": "Sat, 13 Dec 2014 06:39:04 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Hu",
"Yuan-Ting",
""
],
[
"Lin",
"Yen-Yu",
""
],
[
"Chen",
"Hsin-Yi",
""
],
[
"Hsu",
"Kuang-Jui",
""
],
[
"Chen",
"Bing-Yu",
""
]
] | TITLE: Descriptor Ensemble: An Unsupervised Approach to Descriptor Fusion in
the Homography Space
ABSTRACT: With the aim to improve the performance of feature matching, we present an
unsupervised approach to fuse various local descriptors in the space of
homographies. Inspired by the observation that the homographies of correct
feature correspondences vary smoothly along the spatial domain, our approach
stands on the unsupervised nature of feature matching, and can select a good
descriptor for matching each feature point. Specifically, the homography space
serves as the common domain, in which a correspondence obtained by any
descriptor is considered as a point, for integrating various heterogeneous
descriptors. Both geometric coherence and spatial continuity among
correspondences are considered via computing their geodesic distances in the
space. In this way, mutual verification across different descriptors is
allowed, and correct correspondences will be highlighted with a high degree of
consistency (i.e., short geodesic distances here). It follows that one-class
SVM can be applied to identifying these correct correspondences, and boosts the
performance of feature matching. The proposed approach is comprehensively
compared with the state-of-the-art approaches, and evaluated on four benchmarks
of image matching. The promising results manifest its effectiveness.
|
1412.4378 | Bharath Kumar Samanthula | Bharath K. Samanthula, Fang-Yu Rao, Elisa Bertino, Xun Yi, Dongxi Liu | Privacy-Preserving and Outsourced Multi-User k-Means Clustering | 16 pages, 2 figures, 5 tables | null | null | null | cs.CR cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many techniques for privacy-preserving data mining (PPDM) have been
investigated over the past decade. Often, the entities involved in the data
mining process are end-users or organizations with limited computing and
storage resources. As a result, such entities may want to refrain from
participating in the PPDM process. To overcome this issue and to take many
other benefits of cloud computing, outsourcing PPDM tasks to the cloud
environment has recently gained special attention. We consider the scenario
where n entities outsource their databases (in encrypted format) to the cloud
and ask the cloud to perform the clustering task on their combined data in a
privacy-preserving manner. We term such a process as privacy-preserving and
outsourced distributed clustering (PPODC). In this paper, we propose a novel
and efficient solution to the PPODC problem based on k-means clustering
algorithm. The main novelty of our solution lies in avoiding the secure
division operations required in computing cluster centers altogether through an
efficient transformation technique. Our solution builds the clusters securely
in an iterative fashion and returns the final cluster centers to all entities
when a pre-determined termination condition holds. The proposed solution
protects data confidentiality of all the participating entities under the
standard semi-honest model. To the best of our knowledge, ours is the first
work to discuss and propose a comprehensive solution to the PPODC problem that
incurs negligible cost on the participating entities. We theoretically estimate
both the computation and communication costs of the proposed protocol and also
demonstrate its practical value through experiments on a real dataset.
| [
{
"version": "v1",
"created": "Sun, 14 Dec 2014 16:54:26 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Samanthula",
"Bharath K.",
""
],
[
"Rao",
"Fang-Yu",
""
],
[
"Bertino",
"Elisa",
""
],
[
"Yi",
"Xun",
""
],
[
"Liu",
"Dongxi",
""
]
] | TITLE: Privacy-Preserving and Outsourced Multi-User k-Means Clustering
ABSTRACT: Many techniques for privacy-preserving data mining (PPDM) have been
investigated over the past decade. Often, the entities involved in the data
mining process are end-users or organizations with limited computing and
storage resources. As a result, such entities may want to refrain from
participating in the PPDM process. To overcome this issue and to take many
other benefits of cloud computing, outsourcing PPDM tasks to the cloud
environment has recently gained special attention. We consider the scenario
where n entities outsource their databases (in encrypted format) to the cloud
and ask the cloud to perform the clustering task on their combined data in a
privacy-preserving manner. We term such a process as privacy-preserving and
outsourced distributed clustering (PPODC). In this paper, we propose a novel
and efficient solution to the PPODC problem based on k-means clustering
algorithm. The main novelty of our solution lies in avoiding the secure
division operations required in computing cluster centers altogether through an
efficient transformation technique. Our solution builds the clusters securely
in an iterative fashion and returns the final cluster centers to all entities
when a pre-determined termination condition holds. The proposed solution
protects data confidentiality of all the participating entities under the
standard semi-honest model. To the best of our knowledge, ours is the first
work to discuss and propose a comprehensive solution to the PPODC problem that
incurs negligible cost on the participating entities. We theoretically estimate
both the computation and communication costs of the proposed protocol and also
demonstrate its practical value through experiments on a real dataset.
|
1412.4682 | Mykola Pechenizkiy | Erik Tromp and Mykola Pechenizkiy | Rule-based Emotion Detection on Social Media: Putting Tweets on
Plutchik's Wheel | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study sentiment analysis beyond the typical granularity of polarity and
instead use Plutchik's wheel of emotions model. We introduce RBEM-Emo as an
extension to the Rule-Based Emission Model algorithm to deduce such emotions
from human-written messages. We evaluate our approach on two different datasets
and compare its performance with the current state-of-the-art techniques for
emotion detection, including a recursive auto-encoder. The results of the
experimental study suggest that RBEM-Emo is a promising approach advancing the
current state-of-the-art in emotion detection.
| [
{
"version": "v1",
"created": "Mon, 15 Dec 2014 17:20:47 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Tromp",
"Erik",
""
],
[
"Pechenizkiy",
"Mykola",
""
]
] | TITLE: Rule-based Emotion Detection on Social Media: Putting Tweets on
Plutchik's Wheel
ABSTRACT: We study sentiment analysis beyond the typical granularity of polarity and
instead use Plutchik's wheel of emotions model. We introduce RBEM-Emo as an
extension to the Rule-Based Emission Model algorithm to deduce such emotions
from human-written messages. We evaluate our approach on two different datasets
and compare its performance with the current state-of-the-art techniques for
emotion detection, including a recursive auto-encoder. The results of the
experimental study suggest that RBEM-Emo is a promising approach advancing the
current state-of-the-art in emotion detection.
|
1412.4726 | Rustam Tagiew | Rustam Tagiew and Dmitry I. Ignatov and Fadi Amroush | Experimental economics for web mining | 3 pages, 2 tables | null | null | null | cs.CE cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper offers a step towards research infrastructure, which makes data
from experimental economics efficiently usable for analysis of web data. We
believe that regularities of human behavior found in experimental data also
emerge in real world web data. A format for data from experiments is suggested,
which enables its publication as open data. Once standardized datasets of
experiments are available on-line, web mining can take advantages from this
data. Further, the questions about the order of causalities arisen from web
data analysis can inspire new experiment setups.
| [
{
"version": "v1",
"created": "Mon, 15 Dec 2014 19:09:48 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Tagiew",
"Rustam",
""
],
[
"Ignatov",
"Dmitry I.",
""
],
[
"Amroush",
"Fadi",
""
]
] | TITLE: Experimental economics for web mining
ABSTRACT: This paper offers a step towards research infrastructure, which makes data
from experimental economics efficiently usable for analysis of web data. We
believe that regularities of human behavior found in experimental data also
emerge in real world web data. A format for data from experiments is suggested,
which enables its publication as open data. Once standardized datasets of
experiments are available on-line, web mining can take advantages from this
data. Further, the questions about the order of causalities arisen from web
data analysis can inspire new experiment setups.
|
1412.4754 | Yuxiao Dong | Yuxiao Dong, Reid A. Johnson, Nitesh V. Chawla | Will This Paper Increase Your h-index? Scientific Impact Prediction | Proc. of the 8th ACM International Conference on Web Search and Data
Mining (WSDM'15) | null | 10.1145/2684822.2685314 | null | cs.SI cs.DL physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Scientific impact plays a central role in the evaluation of the output of
scholars, departments, and institutions. A widely used measure of scientific
impact is citations, with a growing body of literature focused on predicting
the number of citations obtained by any given publication. The effectiveness of
such predictions, however, is fundamentally limited by the power-law
distribution of citations, whereby publications with few citations are
extremely common and publications with many citations are relatively rare.
Given this limitation, in this work we instead address a related question asked
by many academic researchers in the course of writing a paper, namely: "Will
this paper increase my h-index?" Using a real academic dataset with over 1.7
million authors, 2 million papers, and 8 million citation relationships from
the premier online academic service ArnetMiner, we formalize a novel scientific
impact prediction problem to examine several factors that can drive a paper to
increase the primary author's h-index. We find that the researcher's authority
on the publication topic and the venue in which the paper is published are
crucial factors to the increase of the primary author's h-index, while the
topic popularity and the co-authors' h-indices are of surprisingly little
relevance. By leveraging relevant factors, we find a greater than 87.5%
potential predictability for whether a paper will contribute to an author's
h-index within five years. As a further experiment, we generate a
self-prediction for this paper, estimating that there is a 76% probability that
it will contribute to the h-index of the co-author with the highest current
h-index in five years. We conclude that our findings on the quantification of
scientific impact can help researchers to expand their influence and more
effectively leverage their position of "standing on the shoulders of giants."
| [
{
"version": "v1",
"created": "Mon, 15 Dec 2014 20:36:00 GMT"
}
] | 2014-12-16T00:00:00 | [
[
"Dong",
"Yuxiao",
""
],
[
"Johnson",
"Reid A.",
""
],
[
"Chawla",
"Nitesh V.",
""
]
] | TITLE: Will This Paper Increase Your h-index? Scientific Impact Prediction
ABSTRACT: Scientific impact plays a central role in the evaluation of the output of
scholars, departments, and institutions. A widely used measure of scientific
impact is citations, with a growing body of literature focused on predicting
the number of citations obtained by any given publication. The effectiveness of
such predictions, however, is fundamentally limited by the power-law
distribution of citations, whereby publications with few citations are
extremely common and publications with many citations are relatively rare.
Given this limitation, in this work we instead address a related question asked
by many academic researchers in the course of writing a paper, namely: "Will
this paper increase my h-index?" Using a real academic dataset with over 1.7
million authors, 2 million papers, and 8 million citation relationships from
the premier online academic service ArnetMiner, we formalize a novel scientific
impact prediction problem to examine several factors that can drive a paper to
increase the primary author's h-index. We find that the researcher's authority
on the publication topic and the venue in which the paper is published are
crucial factors to the increase of the primary author's h-index, while the
topic popularity and the co-authors' h-indices are of surprisingly little
relevance. By leveraging relevant factors, we find a greater than 87.5%
potential predictability for whether a paper will contribute to an author's
h-index within five years. As a further experiment, we generate a
self-prediction for this paper, estimating that there is a 76% probability that
it will contribute to the h-index of the co-author with the highest current
h-index in five years. We conclude that our findings on the quantification of
scientific impact can help researchers to expand their influence and more
effectively leverage their position of "standing on the shoulders of giants."
|
1412.3898 | Lu Yu | Lu Yu and Junming Huang and Chuang Liu and Zike Zhang | ILCR: Item-based Latent Factors for Sparse Collaborative Retrieval | 10 pages, conference | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Interactions between search and recommendation have recently attracted
significant attention, and several studies have shown that many potential
applications involve with a joint problem of producing recommendations to users
with respect to a given query, termed $Collaborative$ $Retrieval$ (CR).
Successful algorithms designed for CR should be potentially flexible at dealing
with the sparsity challenges since the setup of collaborative retrieval
associates with a given $query$ $\times$ $user$ $\times$ $item$ tensor instead
of traditional $user$ $\times$ $item$ matrix. Recently, several works are
proposed to study CR task from users' perspective. In this paper, we aim to
sufficiently explore the sophisticated relationship of each $query$ $\times$
$user$ $\times$ $item$ triple from items' perspective. By integrating
item-based collaborative information for this joint task, we present an
alternative factorized model that could better evaluate the ranks of those
items with sparse information for the given query-user pair. In addition, we
suggest to employ a recently proposed scalable ranking learning algorithm,
namely BPR, to optimize the state-of-the-art approach, $Latent$ $Collaborative$
$Retrieval$ model, instead of the original learning algorithm. The experimental
results on two real-world datasets, (i.e. \emph{Last.fm}, \emph{Yelp}),
demonstrate the efficiency and effectiveness of our proposed approach.
| [
{
"version": "v1",
"created": "Fri, 12 Dec 2014 06:32:47 GMT"
}
] | 2014-12-15T00:00:00 | [
[
"Yu",
"Lu",
""
],
[
"Huang",
"Junming",
""
],
[
"Liu",
"Chuang",
""
],
[
"Zhang",
"Zike",
""
]
] | TITLE: ILCR: Item-based Latent Factors for Sparse Collaborative Retrieval
ABSTRACT: Interactions between search and recommendation have recently attracted
significant attention, and several studies have shown that many potential
applications involve with a joint problem of producing recommendations to users
with respect to a given query, termed $Collaborative$ $Retrieval$ (CR).
Successful algorithms designed for CR should be potentially flexible at dealing
with the sparsity challenges since the setup of collaborative retrieval
associates with a given $query$ $\times$ $user$ $\times$ $item$ tensor instead
of traditional $user$ $\times$ $item$ matrix. Recently, several works are
proposed to study CR task from users' perspective. In this paper, we aim to
sufficiently explore the sophisticated relationship of each $query$ $\times$
$user$ $\times$ $item$ triple from items' perspective. By integrating
item-based collaborative information for this joint task, we present an
alternative factorized model that could better evaluate the ranks of those
items with sparse information for the given query-user pair. In addition, we
suggest to employ a recently proposed scalable ranking learning algorithm,
namely BPR, to optimize the state-of-the-art approach, $Latent$ $Collaborative$
$Retrieval$ model, instead of the original learning algorithm. The experimental
results on two real-world datasets, (i.e. \emph{Last.fm}, \emph{Yelp}),
demonstrate the efficiency and effectiveness of our proposed approach.
|
1412.3919 | Alexandre Abraham | Alexandre Abraham (NEUROSPIN, INRIA Saclay - Ile de France), Fabian
Pedregosa (INRIA Saclay - Ile de France), Michael Eickenberg (LNAO, INRIA
Saclay - Ile de France), Philippe Gervais (NEUROSPIN, INRIA Saclay - Ile de
France, LNAO), Andreas Muller, Jean Kossaifi, Alexandre Gramfort (NEUROSPIN,
LTCI), Bertrand Thirion (NEUROSPIN, INRIA Saclay - Ile de France), G\"ael
Varoquaux (NEUROSPIN, INRIA Saclay - Ile de France, LNAO) | Machine Learning for Neuroimaging with Scikit-Learn | Frontiers in neuroscience, Frontiers Research Foundation, 2013, pp.15 | null | null | null | cs.LG cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Statistical machine learning methods are increasingly used for neuroimaging
data analysis. Their main virtue is their ability to model high-dimensional
datasets, e.g. multivariate analysis of activation images or resting-state time
series. Supervised learning is typically used in decoding or encoding settings
to relate brain images to behavioral or clinical observations, while
unsupervised learning can uncover hidden structures in sets of images (e.g.
resting state functional MRI) or find sub-populations in large cohorts. By
considering different functional neuroimaging applications, we illustrate how
scikit-learn, a Python machine learning library, can be used to perform some
key analysis steps. Scikit-learn contains a very large set of statistical
learning algorithms, both supervised and unsupervised, and its application to
neuroimaging data provides a versatile tool to study the brain.
| [
{
"version": "v1",
"created": "Fri, 12 Dec 2014 08:38:35 GMT"
}
] | 2014-12-15T00:00:00 | [
[
"Abraham",
"Alexandre",
"",
"NEUROSPIN, INRIA Saclay - Ile de France"
],
[
"Pedregosa",
"Fabian",
"",
"INRIA Saclay - Ile de France"
],
[
"Eickenberg",
"Michael",
"",
"LNAO, INRIA\n Saclay - Ile de France"
],
[
"Gervais",
"Philippe",
"",
"NEUROSPIN, INRIA Saclay - Ile de\n France, LNAO"
],
[
"Muller",
"Andreas",
"",
"NEUROSPIN,\n LTCI"
],
[
"Kossaifi",
"Jean",
"",
"NEUROSPIN,\n LTCI"
],
[
"Gramfort",
"Alexandre",
"",
"NEUROSPIN,\n LTCI"
],
[
"Thirion",
"Bertrand",
"",
"NEUROSPIN, INRIA Saclay - Ile de France"
],
[
"Varoquaux",
"Gäel",
"",
"NEUROSPIN, INRIA Saclay - Ile de France, LNAO"
]
] | TITLE: Machine Learning for Neuroimaging with Scikit-Learn
ABSTRACT: Statistical machine learning methods are increasingly used for neuroimaging
data analysis. Their main virtue is their ability to model high-dimensional
datasets, e.g. multivariate analysis of activation images or resting-state time
series. Supervised learning is typically used in decoding or encoding settings
to relate brain images to behavioral or clinical observations, while
unsupervised learning can uncover hidden structures in sets of images (e.g.
resting state functional MRI) or find sub-populations in large cohorts. By
considering different functional neuroimaging applications, we illustrate how
scikit-learn, a Python machine learning library, can be used to perform some
key analysis steps. Scikit-learn contains a very large set of statistical
learning algorithms, both supervised and unsupervised, and its application to
neuroimaging data provides a versatile tool to study the brain.
|
1412.4042 | D\'aniel Kondor Mr | D\'aniel Kondor, Istv\'an Csabai, J\'anos Sz\"ule, M\'arton P\'osfai,
G\'abor Vattay | Inferring the interplay of network structure and market effects in
Bitcoin | project website: http://www.vo.elte.hu/bitcoin | New J. Phys. 16 (2014) 125003 | 10.1088/1367-2630/16/12/125003 | null | physics.soc-ph cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A main focus in economics research is understanding the time series of prices
of goods and assets. While statistical models using only the properties of the
time series itself have been successful in many aspects, we expect to gain a
better understanding of the phenomena involved if we can model the underlying
system of interacting agents. In this article, we consider the history of
Bitcoin, a novel digital currency system, for which the complete list of
transactions is available for analysis. Using this dataset, we reconstruct the
transaction network between users and analyze changes in the structure of the
subgraph induced by the most active users. Our approach is based on the
unsupervised identification of important features of the time variation of the
network. Applying the widely used method of Principal Component Analysis to the
matrix constructed from snapshots of the network at different times, we are
able to show how structural changes in the network accompany significant
changes in the exchange price of bitcoins.
| [
{
"version": "v1",
"created": "Fri, 12 Dec 2014 16:31:24 GMT"
}
] | 2014-12-15T00:00:00 | [
[
"Kondor",
"Dániel",
""
],
[
"Csabai",
"István",
""
],
[
"Szüle",
"János",
""
],
[
"Pósfai",
"Márton",
""
],
[
"Vattay",
"Gábor",
""
]
] | TITLE: Inferring the interplay of network structure and market effects in
Bitcoin
ABSTRACT: A main focus in economics research is understanding the time series of prices
of goods and assets. While statistical models using only the properties of the
time series itself have been successful in many aspects, we expect to gain a
better understanding of the phenomena involved if we can model the underlying
system of interacting agents. In this article, we consider the history of
Bitcoin, a novel digital currency system, for which the complete list of
transactions is available for analysis. Using this dataset, we reconstruct the
transaction network between users and analyze changes in the structure of the
subgraph induced by the most active users. Our approach is based on the
unsupervised identification of important features of the time variation of the
network. Applying the widely used method of Principal Component Analysis to the
matrix constructed from snapshots of the network at different times, we are
able to show how structural changes in the network accompany significant
changes in the exchange price of bitcoins.
|
1412.4102 | Chunyu Wang | Chunyu Wang, John Flynn, Yizhou Wang, Alan L. Yuille | Representing Data by a Mixture of Activated Simplices | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a new model which represents data as a mixture of simplices.
Simplices are geometric structures that generalize triangles. We give a simple
geometric understanding that allows us to learn a simplicial structure
efficiently. Our method requires that the data are unit normalized (and thus
lie on the unit sphere). We show that under this restriction, building a model
with simplices amounts to constructing a convex hull inside the sphere whose
boundary facets is close to the data. We call the boundary facets of the convex
hull that are close to the data Activated Simplices. While the total number of
bases used to build the simplices is a parameter of the model, the dimensions
of the individual activated simplices are learned from the data. Simplices can
have different dimensions, which facilitates modeling of inhomogeneous data
sources. The simplicial structure is bounded --- this is appropriate for
modeling data with constraints, such as human elbows can not bend more than 180
degrees. The simplices are easy to interpret and extremes within the data can
be discovered among the vertices. The method provides good reconstruction and
regularization. It supports good nearest neighbor classification and it allows
realistic generative models to be constructed. It achieves state-of-the-art
results on benchmark datasets, including 3D poses and digits.
| [
{
"version": "v1",
"created": "Fri, 12 Dec 2014 20:12:40 GMT"
}
] | 2014-12-15T00:00:00 | [
[
"Wang",
"Chunyu",
""
],
[
"Flynn",
"John",
""
],
[
"Wang",
"Yizhou",
""
],
[
"Yuille",
"Alan L.",
""
]
] | TITLE: Representing Data by a Mixture of Activated Simplices
ABSTRACT: We present a new model which represents data as a mixture of simplices.
Simplices are geometric structures that generalize triangles. We give a simple
geometric understanding that allows us to learn a simplicial structure
efficiently. Our method requires that the data are unit normalized (and thus
lie on the unit sphere). We show that under this restriction, building a model
with simplices amounts to constructing a convex hull inside the sphere whose
boundary facets is close to the data. We call the boundary facets of the convex
hull that are close to the data Activated Simplices. While the total number of
bases used to build the simplices is a parameter of the model, the dimensions
of the individual activated simplices are learned from the data. Simplices can
have different dimensions, which facilitates modeling of inhomogeneous data
sources. The simplicial structure is bounded --- this is appropriate for
modeling data with constraints, such as human elbows can not bend more than 180
degrees. The simplices are easy to interpret and extremes within the data can
be discovered among the vertices. The method provides good reconstruction and
regularization. It supports good nearest neighbor classification and it allows
realistic generative models to be constructed. It achieves state-of-the-art
results on benchmark datasets, including 3D poses and digits.
|
1309.3421 | Yanshan Wang | Yanshan Wang, Jae-Sung Lee, In-Chan Choi | Indexing by Latent Dirichlet Allocation and Ensemble Model | Accepted by the Journal of American Society for Information Science
and Technology (JASIST) | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The contribution of this paper is two-fold. First, we present Indexing by
Latent Dirichlet Allocation (LDI), an automatic document indexing method. The
probability distributions in LDI utilize those in Latent Dirichlet Allocation
(LDA), a generative topic model that has been previously used in applications
for document retrieval tasks. However, the ad hoc applications, or their
variants with smoothing techniques as prompted by previous studies in LDA-based
language modeling, result in unsatisfactory performance as the document
representations do not accurately reflect concept space. To improve
performance, we introduce a new definition of document probability vectors in
the context of LDA and present a novel scheme for automatic document indexing
based on LDA. Second, we propose an Ensemble Model (EnM) for document
retrieval. The EnM combines basis indexing models by assigning different
weights and attempts to uncover the optimal weights to maximize the Mean
Average Precision (MAP). To solve the optimization problem, we propose an
algorithm, EnM.B, which is derived based on the boosting method. The results of
our computational experiments on benchmark data sets indicate that both the
proposed approaches are viable options for document retrieval.
| [
{
"version": "v1",
"created": "Fri, 13 Sep 2013 10:37:47 GMT"
},
{
"version": "v2",
"created": "Thu, 28 Nov 2013 11:37:58 GMT"
},
{
"version": "v3",
"created": "Tue, 14 Oct 2014 07:41:46 GMT"
},
{
"version": "v4",
"created": "Tue, 21 Oct 2014 12:57:14 GMT"
},
{
"version": "v5",
"created": "Mon, 8 Dec 2014 08:27:49 GMT"
},
{
"version": "v6",
"created": "Thu, 11 Dec 2014 05:18:32 GMT"
}
] | 2014-12-12T00:00:00 | [
[
"Wang",
"Yanshan",
""
],
[
"Lee",
"Jae-Sung",
""
],
[
"Choi",
"In-Chan",
""
]
] | TITLE: Indexing by Latent Dirichlet Allocation and Ensemble Model
ABSTRACT: The contribution of this paper is two-fold. First, we present Indexing by
Latent Dirichlet Allocation (LDI), an automatic document indexing method. The
probability distributions in LDI utilize those in Latent Dirichlet Allocation
(LDA), a generative topic model that has been previously used in applications
for document retrieval tasks. However, the ad hoc applications, or their
variants with smoothing techniques as prompted by previous studies in LDA-based
language modeling, result in unsatisfactory performance as the document
representations do not accurately reflect concept space. To improve
performance, we introduce a new definition of document probability vectors in
the context of LDA and present a novel scheme for automatic document indexing
based on LDA. Second, we propose an Ensemble Model (EnM) for document
retrieval. The EnM combines basis indexing models by assigning different
weights and attempts to uncover the optimal weights to maximize the Mean
Average Precision (MAP). To solve the optimization problem, we propose an
algorithm, EnM.B, which is derived based on the boosting method. The results of
our computational experiments on benchmark data sets indicate that both the
proposed approaches are viable options for document retrieval.
|
1406.0146 | Darko Hric | Darko Hric, Richard K. Darst, Santo Fortunato | Community detection in networks: Structural communities versus ground
truth | 21 pages, 19 figures | Phys. Rev. E 90, 062805 (2014) | 10.1103/PhysRevE.90.062805 | null | physics.soc-ph cs.IR cs.SI q-bio.QM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Algorithms to find communities in networks rely just on structural
information and search for cohesive subsets of nodes. On the other hand, most
scholars implicitly or explicitly assume that structural communities represent
groups of nodes with similar (non-topological) properties or functions. This
hypothesis could not be verified, so far, because of the lack of network
datasets with information on the classification of the nodes. We show that
traditional community detection methods fail to find the metadata groups in
many large networks. Our results show that there is a marked separation between
structural communities and metadata groups, in line with recent findings. That
means that either our current modeling of community structure has to be
substantially modified, or that metadata groups may not be recoverable from
topology alone.
| [
{
"version": "v1",
"created": "Sun, 1 Jun 2014 09:06:16 GMT"
},
{
"version": "v2",
"created": "Thu, 11 Dec 2014 18:08:15 GMT"
}
] | 2014-12-12T00:00:00 | [
[
"Hric",
"Darko",
""
],
[
"Darst",
"Richard K.",
""
],
[
"Fortunato",
"Santo",
""
]
] | TITLE: Community detection in networks: Structural communities versus ground
truth
ABSTRACT: Algorithms to find communities in networks rely just on structural
information and search for cohesive subsets of nodes. On the other hand, most
scholars implicitly or explicitly assume that structural communities represent
groups of nodes with similar (non-topological) properties or functions. This
hypothesis could not be verified, so far, because of the lack of network
datasets with information on the classification of the nodes. We show that
traditional community detection methods fail to find the metadata groups in
many large networks. Our results show that there is a marked separation between
structural communities and metadata groups, in line with recent findings. That
means that either our current modeling of community structure has to be
substantially modified, or that metadata groups may not be recoverable from
topology alone.
|
1412.3474 | Eric Tzeng | Eric Tzeng, Judy Hoffman, Ning Zhang, Kate Saenko, Trevor Darrell | Deep Domain Confusion: Maximizing for Domain Invariance | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent reports suggest that a generic supervised deep CNN model trained on a
large-scale dataset reduces, but does not remove, dataset bias on a standard
benchmark. Fine-tuning deep models in a new domain can require a significant
amount of data, which for many applications is simply not available. We propose
a new CNN architecture which introduces an adaptation layer and an additional
domain confusion loss, to learn a representation that is both semantically
meaningful and domain invariant. We additionally show that a domain confusion
metric can be used for model selection to determine the dimension of an
adaptation layer and the best position for the layer in the CNN architecture.
Our proposed adaptation method offers empirical performance which exceeds
previously published results on a standard benchmark visual domain adaptation
task.
| [
{
"version": "v1",
"created": "Wed, 10 Dec 2014 21:20:54 GMT"
}
] | 2014-12-12T00:00:00 | [
[
"Tzeng",
"Eric",
""
],
[
"Hoffman",
"Judy",
""
],
[
"Zhang",
"Ning",
""
],
[
"Saenko",
"Kate",
""
],
[
"Darrell",
"Trevor",
""
]
] | TITLE: Deep Domain Confusion: Maximizing for Domain Invariance
ABSTRACT: Recent reports suggest that a generic supervised deep CNN model trained on a
large-scale dataset reduces, but does not remove, dataset bias on a standard
benchmark. Fine-tuning deep models in a new domain can require a significant
amount of data, which for many applications is simply not available. We propose
a new CNN architecture which introduces an adaptation layer and an additional
domain confusion loss, to learn a representation that is both semantically
meaningful and domain invariant. We additionally show that a domain confusion
metric can be used for model selection to determine the dimension of an
adaptation layer and the best position for the layer in the CNN architecture.
Our proposed adaptation method offers empirical performance which exceeds
previously published results on a standard benchmark visual domain adaptation
task.
|
1412.3684 | Soren Goyal | Soren Goyal, Paul Benjamin | Object Recognition Using Deep Neural Networks: A Survey | null | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recognition of objects using Deep Neural Networks is an active area of
research and many breakthroughs have been made in the last few years. The paper
attempts to indicate how far this field has progressed. The paper briefly
describes the history of research in Neural Networks and describe several of
the recent advances in this field. The performances of recently developed
Neural Network Algorithm over benchmark datasets have been tabulated. Finally,
some the applications of this field have been provided.
| [
{
"version": "v1",
"created": "Wed, 10 Dec 2014 18:23:13 GMT"
}
] | 2014-12-12T00:00:00 | [
[
"Goyal",
"Soren",
""
],
[
"Benjamin",
"Paul",
""
]
] | TITLE: Object Recognition Using Deep Neural Networks: A Survey
ABSTRACT: Recognition of objects using Deep Neural Networks is an active area of
research and many breakthroughs have been made in the last few years. The paper
attempts to indicate how far this field has progressed. The paper briefly
describes the history of research in Neural Networks and describe several of
the recent advances in this field. The performances of recently developed
Neural Network Algorithm over benchmark datasets have been tabulated. Finally,
some the applications of this field have been provided.
|
1402.5450 | Nicholas Rotella | Nicholas Rotella, Michael Bloesch, Ludovic Righetti and Stefan Schaal | State Estimation for a Humanoid Robot | IROS 2014 Submission, IEEE/RSJ International Conference on
Intelligent Robots and Systems (2014) 952-958 | null | 10.1109/IROS.2014.6942674 | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces a framework for state estimation on a humanoid robot
platform using only common proprioceptive sensors and knowledge of leg
kinematics. The presented approach extends that detailed in [1] on a quadruped
platform by incorporating the rotational constraints imposed by the humanoid's
flat feet. As in previous work, the proposed Extended Kalman Filter (EKF)
accommodates contact switching and makes no assumptions about gait or terrain,
making it applicable on any humanoid platform for use in any task. The filter
employs a sensor-based prediction model which uses inertial data from an IMU
and corrects for integrated error using a kinematics-based measurement model
which relies on joint encoders and a kinematic model to determine the relative
position and orientation of the feet. A nonlinear observability analysis is
performed on both the original and updated filters and it is concluded that the
new filter significantly simplifies singular cases and improves the
observability characteristics of the system. Results on simulated walking and
squatting datasets demonstrate the performance gain of the flat-foot filter as
well as confirm the results of the presented observability analysis.
| [
{
"version": "v1",
"created": "Fri, 21 Feb 2014 23:35:34 GMT"
},
{
"version": "v2",
"created": "Wed, 10 Dec 2014 20:42:58 GMT"
}
] | 2014-12-11T00:00:00 | [
[
"Rotella",
"Nicholas",
""
],
[
"Bloesch",
"Michael",
""
],
[
"Righetti",
"Ludovic",
""
],
[
"Schaal",
"Stefan",
""
]
] | TITLE: State Estimation for a Humanoid Robot
ABSTRACT: This paper introduces a framework for state estimation on a humanoid robot
platform using only common proprioceptive sensors and knowledge of leg
kinematics. The presented approach extends that detailed in [1] on a quadruped
platform by incorporating the rotational constraints imposed by the humanoid's
flat feet. As in previous work, the proposed Extended Kalman Filter (EKF)
accommodates contact switching and makes no assumptions about gait or terrain,
making it applicable on any humanoid platform for use in any task. The filter
employs a sensor-based prediction model which uses inertial data from an IMU
and corrects for integrated error using a kinematics-based measurement model
which relies on joint encoders and a kinematic model to determine the relative
position and orientation of the feet. A nonlinear observability analysis is
performed on both the original and updated filters and it is concluded that the
new filter significantly simplifies singular cases and improves the
observability characteristics of the system. Results on simulated walking and
squatting datasets demonstrate the performance gain of the flat-foot filter as
well as confirm the results of the presented observability analysis.
|
1412.3161 | Xiaoyu Wang | Xiaoyu Wang, Tianbao Yang, Guobin Chen, Yuanqing Lin | Object-centric Sampling for Fine-grained Image Classification | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes to go beyond the state-of-the-art deep convolutional
neural network (CNN) by incorporating the information from object detection,
focusing on dealing with fine-grained image classification. Unfortunately, CNN
suffers from over-fiting when it is trained on existing fine-grained image
classification benchmarks, which typically only consist of less than a few tens
of thousands training images. Therefore, we first construct a large-scale
fine-grained car recognition dataset that consists of 333 car classes with more
than 150 thousand training images. With this large-scale dataset, we are able
to build a strong baseline for CNN with top-1 classification accuracy of 81.6%.
One major challenge in fine-grained image classification is that many classes
are very similar to each other while having large within-class variation. One
contributing factor to the within-class variation is cluttered image
background. However, the existing CNN training takes uniform window sampling
over the image, acting as blind on the location of the object of interest. In
contrast, this paper proposes an \emph{object-centric sampling} (OCS) scheme
that samples image windows based on the object location information. The
challenge in using the location information lies in how to design powerful
object detector and how to handle the imperfectness of detection results. To
that end, we design a saliency-aware object detection approach specific for the
setting of fine-grained image classification, and the uncertainty of detection
results are naturally handled in our OCS scheme. Our framework is demonstrated
to be very effective, improving top-1 accuracy to 89.3% (from 81.6%) on the
large-scale fine-grained car classification dataset.
| [
{
"version": "v1",
"created": "Wed, 10 Dec 2014 00:28:49 GMT"
}
] | 2014-12-11T00:00:00 | [
[
"Wang",
"Xiaoyu",
""
],
[
"Yang",
"Tianbao",
""
],
[
"Chen",
"Guobin",
""
],
[
"Lin",
"Yuanqing",
""
]
] | TITLE: Object-centric Sampling for Fine-grained Image Classification
ABSTRACT: This paper proposes to go beyond the state-of-the-art deep convolutional
neural network (CNN) by incorporating the information from object detection,
focusing on dealing with fine-grained image classification. Unfortunately, CNN
suffers from over-fiting when it is trained on existing fine-grained image
classification benchmarks, which typically only consist of less than a few tens
of thousands training images. Therefore, we first construct a large-scale
fine-grained car recognition dataset that consists of 333 car classes with more
than 150 thousand training images. With this large-scale dataset, we are able
to build a strong baseline for CNN with top-1 classification accuracy of 81.6%.
One major challenge in fine-grained image classification is that many classes
are very similar to each other while having large within-class variation. One
contributing factor to the within-class variation is cluttered image
background. However, the existing CNN training takes uniform window sampling
over the image, acting as blind on the location of the object of interest. In
contrast, this paper proposes an \emph{object-centric sampling} (OCS) scheme
that samples image windows based on the object location information. The
challenge in using the location information lies in how to design powerful
object detector and how to handle the imperfectness of detection results. To
that end, we design a saliency-aware object detection approach specific for the
setting of fine-grained image classification, and the uncertainty of detection
results are naturally handled in our OCS scheme. Our framework is demonstrated
to be very effective, improving top-1 accuracy to 89.3% (from 81.6%) on the
large-scale fine-grained car classification dataset.
|
1412.3333 | Patrick Prosser | Frod Prefect, Patrick Prosser | Empirical Algorithmics: draw your own conclusions | arXiv admin note: text overlap with arXiv:1207.4616 | null | null | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In an empirical comparisons of algorithms we might compare run times over a
set of benchmark problems to decide which one is fastest, i.e. an algorithmic
horse race. Ideally we would like to download source code for the algorithms,
compile and then run on our machine. Sometimes code isn't available to download
and sometimes resource isn't available to implement all the algorithms we want
to study. To get round this, published results are rescaled, a technique
endorsed by DIMACS, and those rescaled results included in a new study. This
technique is frequently used when presenting new algorithms for the maximum
clique problem. We demonstrate that this is unsafe, and that if carelessly used
may allow us to draw conflicting conclusions from our empirical study.
| [
{
"version": "v1",
"created": "Wed, 10 Dec 2014 15:11:05 GMT"
}
] | 2014-12-11T00:00:00 | [
[
"Prefect",
"Frod",
""
],
[
"Prosser",
"Patrick",
""
]
] | TITLE: Empirical Algorithmics: draw your own conclusions
ABSTRACT: In an empirical comparisons of algorithms we might compare run times over a
set of benchmark problems to decide which one is fastest, i.e. an algorithmic
horse race. Ideally we would like to download source code for the algorithms,
compile and then run on our machine. Sometimes code isn't available to download
and sometimes resource isn't available to implement all the algorithms we want
to study. To get round this, published results are rescaled, a technique
endorsed by DIMACS, and those rescaled results included in a new study. This
technique is frequently used when presenting new algorithms for the maximum
clique problem. We demonstrate that this is unsafe, and that if carelessly used
may allow us to draw conflicting conclusions from our empirical study.
|
1412.3352 | Neda Pourali | Neda Pourali | Web image annotation by diffusion maps manifold learning algorithm | 11 pages, 8 figures | null | 10.5121/ijfcst.2014.4606 | null | cs.CV cs.IR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic image annotation is one of the most challenging problems in machine
vision areas. The goal of this task is to predict number of keywords
automatically for images captured in real data. Many methods are based on
visual features in order to calculate similarities between image samples. But
the computation cost of these approaches is very high. These methods require
many training samples to be stored in memory. To lessen this burden, a number
of techniques have been developed to reduce the number of features in a
dataset. Manifold learning is a popular approach to nonlinear dimensionality
reduction. In this paper, we investigate Diffusion maps manifold learning
method for web image auto-annotation task. Diffusion maps manifold learning
method is used to reduce the dimension of some visual features. Extensive
experiments and analysis on NUS-WIDE-LITE web image dataset with different
visual features show how this manifold learning dimensionality reduction method
can be applied effectively to image annotation.
| [
{
"version": "v1",
"created": "Mon, 8 Dec 2014 10:38:28 GMT"
}
] | 2014-12-11T00:00:00 | [
[
"Pourali",
"Neda",
""
]
] | TITLE: Web image annotation by diffusion maps manifold learning algorithm
ABSTRACT: Automatic image annotation is one of the most challenging problems in machine
vision areas. The goal of this task is to predict number of keywords
automatically for images captured in real data. Many methods are based on
visual features in order to calculate similarities between image samples. But
the computation cost of these approaches is very high. These methods require
many training samples to be stored in memory. To lessen this burden, a number
of techniques have been developed to reduce the number of features in a
dataset. Manifold learning is a popular approach to nonlinear dimensionality
reduction. In this paper, we investigate Diffusion maps manifold learning
method for web image auto-annotation task. Diffusion maps manifold learning
method is used to reduce the dimension of some visual features. Extensive
experiments and analysis on NUS-WIDE-LITE web image dataset with different
visual features show how this manifold learning dimensionality reduction method
can be applied effectively to image annotation.
|
1406.2227 | Max Jaderberg | Max Jaderberg, Karen Simonyan, Andrea Vedaldi, Andrew Zisserman | Synthetic Data and Artificial Neural Networks for Natural Scene Text
Recognition | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we present a framework for the recognition of natural scene
text. Our framework does not require any human-labelled data, and performs word
recognition on the whole image holistically, departing from the character based
recognition systems of the past. The deep neural network models at the centre
of this framework are trained solely on data produced by a synthetic text
generation engine -- synthetic data that is highly realistic and sufficient to
replace real data, giving us infinite amounts of training data. This excess of
data exposes new possibilities for word recognition models, and here we
consider three models, each one "reading" words in a different way: via 90k-way
dictionary encoding, character sequence encoding, and bag-of-N-grams encoding.
In the scenarios of language based and completely unconstrained text
recognition we greatly improve upon state-of-the-art performance on standard
datasets, using our fast, simple machinery and requiring zero data-acquisition
costs.
| [
{
"version": "v1",
"created": "Mon, 9 Jun 2014 15:53:33 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2014 03:10:35 GMT"
},
{
"version": "v3",
"created": "Mon, 6 Oct 2014 16:08:24 GMT"
},
{
"version": "v4",
"created": "Tue, 9 Dec 2014 11:22:59 GMT"
}
] | 2014-12-10T00:00:00 | [
[
"Jaderberg",
"Max",
""
],
[
"Simonyan",
"Karen",
""
],
[
"Vedaldi",
"Andrea",
""
],
[
"Zisserman",
"Andrew",
""
]
] | TITLE: Synthetic Data and Artificial Neural Networks for Natural Scene Text
Recognition
ABSTRACT: In this work we present a framework for the recognition of natural scene
text. Our framework does not require any human-labelled data, and performs word
recognition on the whole image holistically, departing from the character based
recognition systems of the past. The deep neural network models at the centre
of this framework are trained solely on data produced by a synthetic text
generation engine -- synthetic data that is highly realistic and sufficient to
replace real data, giving us infinite amounts of training data. This excess of
data exposes new possibilities for word recognition models, and here we
consider three models, each one "reading" words in a different way: via 90k-way
dictionary encoding, character sequence encoding, and bag-of-N-grams encoding.
In the scenarios of language based and completely unconstrained text
recognition we greatly improve upon state-of-the-art performance on standard
datasets, using our fast, simple machinery and requiring zero data-acquisition
costs.
|
1410.5772 | Lutz Bornmann Dr. | Lutz Bornmann, Werner Marx | Methods for the generation of normalized citation impact scores in
bibliometrics: Which method best reflects the judgements of experts? | Accepted for publication in the Journal of Informetrics | null | null | null | cs.DL stat.OT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Evaluative bibliometrics compares the citation impact of researchers,
research groups and institutions with each other across time scales and
disciplines. Both factors - discipline and period - have an influence on the
citation count which is independent of the quality of the publication.
Normalizing the citation impact of papers for these two factors started in the
mid-1980s. Since then, a range of different methods have been presented for
producing normalized citation impact scores. The current study uses a data set
of over 50,000 records to test which of the methods so far presented correlate
better with the assessment of papers by peers. The peer assessments come from
F1000Prime - a post-publication peer review system of the biomedical
literature. Of the normalized indicators, the current study involves not only
cited-side indicators, such as the mean normalized citation score, but also
citing-side indicators. As the results show, the correlations of the indicators
with the peer assessments all turn out to be very similar. Since F1000 focuses
on biomedicine, it is important that the results of this study are validated by
other studies based on datasets from other disciplines or (ideally) based on
multi-disciplinary datasets.
| [
{
"version": "v1",
"created": "Mon, 20 Oct 2014 07:57:32 GMT"
},
{
"version": "v2",
"created": "Tue, 9 Dec 2014 10:58:06 GMT"
}
] | 2014-12-10T00:00:00 | [
[
"Bornmann",
"Lutz",
""
],
[
"Marx",
"Werner",
""
]
] | TITLE: Methods for the generation of normalized citation impact scores in
bibliometrics: Which method best reflects the judgements of experts?
ABSTRACT: Evaluative bibliometrics compares the citation impact of researchers,
research groups and institutions with each other across time scales and
disciplines. Both factors - discipline and period - have an influence on the
citation count which is independent of the quality of the publication.
Normalizing the citation impact of papers for these two factors started in the
mid-1980s. Since then, a range of different methods have been presented for
producing normalized citation impact scores. The current study uses a data set
of over 50,000 records to test which of the methods so far presented correlate
better with the assessment of papers by peers. The peer assessments come from
F1000Prime - a post-publication peer review system of the biomedical
literature. Of the normalized indicators, the current study involves not only
cited-side indicators, such as the mean normalized citation score, but also
citing-side indicators. As the results show, the correlations of the indicators
with the peer assessments all turn out to be very similar. Since F1000 focuses
on biomedicine, it is important that the results of this study are validated by
other studies based on datasets from other disciplines or (ideally) based on
multi-disciplinary datasets.
|
1410.7835 | Zhensong Qian | Oliver Schulte, Zhensong Qian, Arthur E. Kirkpatrick, Xiaoqian Yin,
Yan Sun | Fast Learning of Relational Dependency Networks | 17 pages, 2 figures, 3 tables, Accepted as long paper by ILP 2014,
September 14- 16th, Nancy, France. Added the Appendix: Proof of Consistency
Characterization | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A Relational Dependency Network (RDN) is a directed graphical model widely
used for multi-relational data. These networks allow cyclic dependencies,
necessary to represent relational autocorrelations. We describe an approach for
learning both the RDN's structure and its parameters, given an input relational
database: First learn a Bayesian network (BN), then transform the Bayesian
network to an RDN. Thus fast Bayes net learning can provide fast RDN learning.
The BN-to-RDN transform comprises a simple, local adjustment of the Bayes net
structure and a closed-form transform of the Bayes net parameters. This method
can learn an RDN for a dataset with a million tuples in minutes. We empirically
compare our approach to state-of-the art RDN learning methods that use
functional gradient boosting, on five benchmark datasets. Learning RDNs via BNs
scales much better to large datasets than learning RDNs with boosting, and
provides competitive accuracy in predictions.
| [
{
"version": "v1",
"created": "Tue, 28 Oct 2014 23:14:56 GMT"
},
{
"version": "v2",
"created": "Tue, 9 Dec 2014 01:07:36 GMT"
}
] | 2014-12-10T00:00:00 | [
[
"Schulte",
"Oliver",
""
],
[
"Qian",
"Zhensong",
""
],
[
"Kirkpatrick",
"Arthur E.",
""
],
[
"Yin",
"Xiaoqian",
""
],
[
"Sun",
"Yan",
""
]
] | TITLE: Fast Learning of Relational Dependency Networks
ABSTRACT: A Relational Dependency Network (RDN) is a directed graphical model widely
used for multi-relational data. These networks allow cyclic dependencies,
necessary to represent relational autocorrelations. We describe an approach for
learning both the RDN's structure and its parameters, given an input relational
database: First learn a Bayesian network (BN), then transform the Bayesian
network to an RDN. Thus fast Bayes net learning can provide fast RDN learning.
The BN-to-RDN transform comprises a simple, local adjustment of the Bayes net
structure and a closed-form transform of the Bayes net parameters. This method
can learn an RDN for a dataset with a million tuples in minutes. We empirically
compare our approach to state-of-the art RDN learning methods that use
functional gradient boosting, on five benchmark datasets. Learning RDNs via BNs
scales much better to large datasets than learning RDNs with boosting, and
provides competitive accuracy in predictions.
|
1412.2066 | Charless Fowlkes | Shaofei Wang and Charless C. Fowlkes | Learning Multi-target Tracking with Quadratic Object Interactions | null | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We describe a model for multi-target tracking based on associating
collections of candidate detections across frames of a video. In order to model
pairwise interactions between different tracks, such as suppression of
overlapping tracks and contextual cues about co-occurence of different objects,
we augment a standard min-cost flow objective with quadratic terms between
detection variables. We learn the parameters of this model using structured
prediction and a loss function which approximates the multi-target tracking
accuracy. We evaluate two different approaches to finding an optimal set of
tracks under model objective based on an LP relaxation and a novel greedy
extension to dynamic programming that handles pairwise interactions. We find
the greedy algorithm achieves equivalent performance to the LP relaxation while
being 2-7x faster than a commercial solver. The resulting model with learned
parameters outperforms existing methods across several categories on the KITTI
tracking benchmark.
| [
{
"version": "v1",
"created": "Fri, 5 Dec 2014 17:04:35 GMT"
},
{
"version": "v2",
"created": "Tue, 9 Dec 2014 05:25:44 GMT"
}
] | 2014-12-10T00:00:00 | [
[
"Wang",
"Shaofei",
""
],
[
"Fowlkes",
"Charless C.",
""
]
] | TITLE: Learning Multi-target Tracking with Quadratic Object Interactions
ABSTRACT: We describe a model for multi-target tracking based on associating
collections of candidate detections across frames of a video. In order to model
pairwise interactions between different tracks, such as suppression of
overlapping tracks and contextual cues about co-occurence of different objects,
we augment a standard min-cost flow objective with quadratic terms between
detection variables. We learn the parameters of this model using structured
prediction and a loss function which approximates the multi-target tracking
accuracy. We evaluate two different approaches to finding an optimal set of
tracks under model objective based on an LP relaxation and a novel greedy
extension to dynamic programming that handles pairwise interactions. We find
the greedy algorithm achieves equivalent performance to the LP relaxation while
being 2-7x faster than a commercial solver. The resulting model with learned
parameters outperforms existing methods across several categories on the KITTI
tracking benchmark.
|
1412.3040 | Albert Kim | Albert Kim, Eric Blais, Aditya Parameswaran, Piotr Indyk, Sam Madden,
Ronitt Rubinfeld | Rapid Sampling for Visualizations with Ordering Guarantees | Tech Report. 17 pages. Condensed version to appear in VLDB Vol. 8 No.
5 | null | null | null | cs.DB | http://creativecommons.org/licenses/by-nc-sa/3.0/ | Visualizations are frequently used as a means to understand trends and gather
insights from datasets, but often take a long time to generate. In this paper,
we focus on the problem of rapidly generating approximate visualizations while
preserving crucial visual proper- ties of interest to analysts. Our primary
focus will be on sampling algorithms that preserve the visual property of
ordering; our techniques will also apply to some other visual properties. For
instance, our algorithms can be used to generate an approximate visualization
of a bar chart very rapidly, where the comparisons between any two bars are
correct. We formally show that our sampling algorithms are generally applicable
and provably optimal in theory, in that they do not take more samples than
necessary to generate the visualizations with ordering guarantees. They also
work well in practice, correctly ordering output groups while taking orders of
magnitude fewer samples and much less time than conventional sampling schemes.
| [
{
"version": "v1",
"created": "Tue, 9 Dec 2014 18:08:26 GMT"
}
] | 2014-12-10T00:00:00 | [
[
"Kim",
"Albert",
""
],
[
"Blais",
"Eric",
""
],
[
"Parameswaran",
"Aditya",
""
],
[
"Indyk",
"Piotr",
""
],
[
"Madden",
"Sam",
""
],
[
"Rubinfeld",
"Ronitt",
""
]
] | TITLE: Rapid Sampling for Visualizations with Ordering Guarantees
ABSTRACT: Visualizations are frequently used as a means to understand trends and gather
insights from datasets, but often take a long time to generate. In this paper,
we focus on the problem of rapidly generating approximate visualizations while
preserving crucial visual proper- ties of interest to analysts. Our primary
focus will be on sampling algorithms that preserve the visual property of
ordering; our techniques will also apply to some other visual properties. For
instance, our algorithms can be used to generate an approximate visualization
of a bar chart very rapidly, where the comparisons between any two bars are
correct. We formally show that our sampling algorithms are generally applicable
and provably optimal in theory, in that they do not take more samples than
necessary to generate the visualizations with ordering guarantees. They also
work well in practice, correctly ordering output groups while taking orders of
magnitude fewer samples and much less time than conventional sampling schemes.
|
1406.1485 | Li Yao | Tapani Raiko, Li Yao, Kyunghyun Cho and Yoshua Bengio | Iterative Neural Autoregressive Distribution Estimator (NADE-k) | Accepted at Neural Information Processing Systems (NIPS) 2014 | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Training of the neural autoregressive density estimator (NADE) can be viewed
as doing one step of probabilistic inference on missing values in data. We
propose a new model that extends this inference scheme to multiple steps,
arguing that it is easier to learn to improve a reconstruction in $k$ steps
rather than to learn to reconstruct in a single inference step. The proposed
model is an unsupervised building block for deep learning that combines the
desirable properties of NADE and multi-predictive training: (1) Its test
likelihood can be computed analytically, (2) it is easy to generate independent
samples from it, and (3) it uses an inference engine that is a superset of
variational inference for Boltzmann machines. The proposed NADE-k is
competitive with the state-of-the-art in density estimation on the two datasets
tested.
| [
{
"version": "v1",
"created": "Thu, 5 Jun 2014 19:13:51 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2014 11:28:36 GMT"
},
{
"version": "v3",
"created": "Sat, 6 Dec 2014 00:22:00 GMT"
}
] | 2014-12-09T00:00:00 | [
[
"Raiko",
"Tapani",
""
],
[
"Yao",
"Li",
""
],
[
"Cho",
"Kyunghyun",
""
],
[
"Bengio",
"Yoshua",
""
]
] | TITLE: Iterative Neural Autoregressive Distribution Estimator (NADE-k)
ABSTRACT: Training of the neural autoregressive density estimator (NADE) can be viewed
as doing one step of probabilistic inference on missing values in data. We
propose a new model that extends this inference scheme to multiple steps,
arguing that it is easier to learn to improve a reconstruction in $k$ steps
rather than to learn to reconstruct in a single inference step. The proposed
model is an unsupervised building block for deep learning that combines the
desirable properties of NADE and multi-predictive training: (1) Its test
likelihood can be computed analytically, (2) it is easy to generate independent
samples from it, and (3) it uses an inference engine that is a superset of
variational inference for Boltzmann machines. The proposed NADE-k is
competitive with the state-of-the-art in density estimation on the two datasets
tested.
|
1411.1792 | Jason Yosinski | Jason Yosinski, Jeff Clune, Yoshua Bengio, Hod Lipson | How transferable are features in deep neural networks? | To appear in Advances in Neural Information Processing Systems 27
(NIPS 2014) | Advances in Neural Information Processing Systems 27, pages
3320-3328. Dec. 2014 | null | null | cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many deep neural networks trained on natural images exhibit a curious
phenomenon in common: on the first layer they learn features similar to Gabor
filters and color blobs. Such first-layer features appear not to be specific to
a particular dataset or task, but general in that they are applicable to many
datasets and tasks. Features must eventually transition from general to
specific by the last layer of the network, but this transition has not been
studied extensively. In this paper we experimentally quantify the generality
versus specificity of neurons in each layer of a deep convolutional neural
network and report a few surprising results. Transferability is negatively
affected by two distinct issues: (1) the specialization of higher layer neurons
to their original task at the expense of performance on the target task, which
was expected, and (2) optimization difficulties related to splitting networks
between co-adapted neurons, which was not expected. In an example network
trained on ImageNet, we demonstrate that either of these two issues may
dominate, depending on whether features are transferred from the bottom,
middle, or top of the network. We also document that the transferability of
features decreases as the distance between the base task and target task
increases, but that transferring features even from distant tasks can be better
than using random features. A final surprising result is that initializing a
network with transferred features from almost any number of layers can produce
a boost to generalization that lingers even after fine-tuning to the target
dataset.
| [
{
"version": "v1",
"created": "Thu, 6 Nov 2014 23:09:37 GMT"
}
] | 2014-12-09T00:00:00 | [
[
"Yosinski",
"Jason",
""
],
[
"Clune",
"Jeff",
""
],
[
"Bengio",
"Yoshua",
""
],
[
"Lipson",
"Hod",
""
]
] | TITLE: How transferable are features in deep neural networks?
ABSTRACT: Many deep neural networks trained on natural images exhibit a curious
phenomenon in common: on the first layer they learn features similar to Gabor
filters and color blobs. Such first-layer features appear not to be specific to
a particular dataset or task, but general in that they are applicable to many
datasets and tasks. Features must eventually transition from general to
specific by the last layer of the network, but this transition has not been
studied extensively. In this paper we experimentally quantify the generality
versus specificity of neurons in each layer of a deep convolutional neural
network and report a few surprising results. Transferability is negatively
affected by two distinct issues: (1) the specialization of higher layer neurons
to their original task at the expense of performance on the target task, which
was expected, and (2) optimization difficulties related to splitting networks
between co-adapted neurons, which was not expected. In an example network
trained on ImageNet, we demonstrate that either of these two issues may
dominate, depending on whether features are transferred from the bottom,
middle, or top of the network. We also document that the transferability of
features decreases as the distance between the base task and target task
increases, but that transferring features even from distant tasks can be better
than using random features. A final surprising result is that initializing a
network with transferred features from almost any number of layers can produce
a boost to generalization that lingers even after fine-tuning to the target
dataset.
|
1412.2485 | Vikram Nathan | Vikram Nathan, Sharath Raghvendra | Accurate Streaming Support Vector Machines | 2 figures, 8 pages | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A widely-used tool for binary classification is the Support Vector Machine
(SVM), a supervised learning technique that finds the "maximum margin" linear
separator between the two classes. While SVMs have been well studied in the
batch (offline) setting, there is considerably less work on the streaming
(online) setting, which requires only a single pass over the data using
sub-linear space. Existing streaming algorithms are not yet competitive with
the batch implementation. In this paper, we use the formulation of the SVM as a
minimum enclosing ball (MEB) problem to provide a streaming SVM algorithm based
off of the blurred ball cover originally proposed by Agarwal and Sharathkumar.
Our implementation consistently outperforms existing streaming SVM approaches
and provides higher accuracies than libSVM on several datasets, thus making it
competitive with the standard SVM batch implementation.
| [
{
"version": "v1",
"created": "Mon, 8 Dec 2014 08:46:07 GMT"
}
] | 2014-12-09T00:00:00 | [
[
"Nathan",
"Vikram",
""
],
[
"Raghvendra",
"Sharath",
""
]
] | TITLE: Accurate Streaming Support Vector Machines
ABSTRACT: A widely-used tool for binary classification is the Support Vector Machine
(SVM), a supervised learning technique that finds the "maximum margin" linear
separator between the two classes. While SVMs have been well studied in the
batch (offline) setting, there is considerably less work on the streaming
(online) setting, which requires only a single pass over the data using
sub-linear space. Existing streaming algorithms are not yet competitive with
the batch implementation. In this paper, we use the formulation of the SVM as a
minimum enclosing ball (MEB) problem to provide a streaming SVM algorithm based
off of the blurred ball cover originally proposed by Agarwal and Sharathkumar.
Our implementation consistently outperforms existing streaming SVM approaches
and provides higher accuracies than libSVM on several datasets, thus making it
competitive with the standard SVM batch implementation.
|
1412.2697 | Hocine Cherifi | Abdelkaher Ait Abdelouahad, Mohammed El Hassouni, Hocine Cherifi, and
Driss Aboutajdine | Image quality assessment measure based on natural image statistics in
the Tetrolet domain | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper deals with a reduced reference (RR) image quality measure based on
natural image statistics modeling. For this purpose, Tetrolet transform is used
since it provides a convenient way to capture local geometric structures. This
transform is applied to both reference and distorted images. Then, Gaussian
Scale Mixture (GSM) is proposed to model subbands in order to take account
statistical dependencies between tetrolet coefficients. In order to quantify
the visual degradation, a measure based on Kullback Leibler Divergence (KLD) is
provided. The proposed measure was tested on the Cornell VCL A-57 dataset and
compared with other measures according to FR-TV1 VQEG framework.
| [
{
"version": "v1",
"created": "Mon, 8 Dec 2014 18:48:26 GMT"
}
] | 2014-12-09T00:00:00 | [
[
"Abdelouahad",
"Abdelkaher Ait",
""
],
[
"Hassouni",
"Mohammed El",
""
],
[
"Cherifi",
"Hocine",
""
],
[
"Aboutajdine",
"Driss",
""
]
] | TITLE: Image quality assessment measure based on natural image statistics in
the Tetrolet domain
ABSTRACT: This paper deals with a reduced reference (RR) image quality measure based on
natural image statistics modeling. For this purpose, Tetrolet transform is used
since it provides a convenient way to capture local geometric structures. This
transform is applied to both reference and distorted images. Then, Gaussian
Scale Mixture (GSM) is proposed to model subbands in order to take account
statistical dependencies between tetrolet coefficients. In order to quantify
the visual degradation, a measure based on Kullback Leibler Divergence (KLD) is
provided. The proposed measure was tested on the Cornell VCL A-57 dataset and
compared with other measures according to FR-TV1 VQEG framework.
|
1412.2716 | Paul Ginsparg | Daniel T. Citron and Paul Ginsparg | Patterns of Text Reuse in a Scientific Corpus | 6 pages, plus 10 pages of supplementary material. To appear in PNAS
(online 8 Dec 2014) | null | 10.1073/pnas.1415135111 | null | cs.DL physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the incidence of text "reuse" by researchers, via a systematic
pairwise comparison of the text content of all articles deposited to arXiv.org
from 1991--2012. We measure the global frequencies of three classes of text
reuse, and measure how chronic text reuse is distributed among authors in the
dataset. We infer a baseline for accepted practice, perhaps surprisingly
permissive compared with other societal contexts, and a clearly delineated set
of aberrant authors. We find a negative correlation between the amount of
reused text in an article and its influence, as measured by subsequent
citations. Finally, we consider the distribution of countries of origin of
articles containing large amounts of reused text.
| [
{
"version": "v1",
"created": "Mon, 8 Dec 2014 20:01:17 GMT"
}
] | 2014-12-09T00:00:00 | [
[
"Citron",
"Daniel T.",
""
],
[
"Ginsparg",
"Paul",
""
]
] | TITLE: Patterns of Text Reuse in a Scientific Corpus
ABSTRACT: We consider the incidence of text "reuse" by researchers, via a systematic
pairwise comparison of the text content of all articles deposited to arXiv.org
from 1991--2012. We measure the global frequencies of three classes of text
reuse, and measure how chronic text reuse is distributed among authors in the
dataset. We infer a baseline for accepted practice, perhaps surprisingly
permissive compared with other societal contexts, and a clearly delineated set
of aberrant authors. We find a negative correlation between the amount of
reused text in an article and its influence, as measured by subsequent
citations. Finally, we consider the distribution of countries of origin of
articles containing large amounts of reused text.
|
1404.7467 | Fangfang Li | Fangfang Li, Guandong Xu, Longbing Cao | Coupled Matrix Factorization within Non-IID Context | 12 pages submitted to PAKDD 2015 | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recommender systems research has experienced different stages such as from
user preference understanding to content analysis. Typical recommendation
algorithms were built on the following bases: (1) assuming users and items are
IID, namely independent and identically distributed, and (2) focusing on
specific aspects such as user preferences or contents. In reality, complex
recommendation tasks involve and request (1) personalized outcomes to tailor
heterogeneous subjective preferences; and (2) explicit and implicit objective
coupling relationships between users, items, and ratings to be considered as
intrinsic forces driving preferences. This inevitably involves the non-IID
complexity and the need of combining subjective preference with objective
couplings hidden in recommendation applications. In this paper, we propose a
novel generic coupled matrix factorization (CMF) model by incorporating non-IID
coupling relations between users and items. Such couplings integrate the
intra-coupled interactions within an attribute and inter-coupled interactions
among different attributes. Experimental results on two open data sets
demonstrate that the user/item couplings can be effectively applied in RS and
CMF outperforms the benchmark methods.
| [
{
"version": "v1",
"created": "Tue, 8 Apr 2014 02:10:16 GMT"
},
{
"version": "v2",
"created": "Thu, 4 Dec 2014 23:29:33 GMT"
}
] | 2014-12-08T00:00:00 | [
[
"Li",
"Fangfang",
""
],
[
"Xu",
"Guandong",
""
],
[
"Cao",
"Longbing",
""
]
] | TITLE: Coupled Matrix Factorization within Non-IID Context
ABSTRACT: Recommender systems research has experienced different stages such as from
user preference understanding to content analysis. Typical recommendation
algorithms were built on the following bases: (1) assuming users and items are
IID, namely independent and identically distributed, and (2) focusing on
specific aspects such as user preferences or contents. In reality, complex
recommendation tasks involve and request (1) personalized outcomes to tailor
heterogeneous subjective preferences; and (2) explicit and implicit objective
coupling relationships between users, items, and ratings to be considered as
intrinsic forces driving preferences. This inevitably involves the non-IID
complexity and the need of combining subjective preference with objective
couplings hidden in recommendation applications. In this paper, we propose a
novel generic coupled matrix factorization (CMF) model by incorporating non-IID
coupling relations between users and items. Such couplings integrate the
intra-coupled interactions within an attribute and inter-coupled interactions
among different attributes. Experimental results on two open data sets
demonstrate that the user/item couplings can be effectively applied in RS and
CMF outperforms the benchmark methods.
|
1410.7454 | Neeraj Dhungel | Neeraj Dhungel, Gustavo Carneiro, Andrew P. Bradley | Deep Structured learning for mass segmentation from Mammograms | 4 pages, 2 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by/3.0/ | In this paper, we present a novel method for the segmentation of breast
masses from mammograms exploring structured and deep learning. Specifically,
using structured support vector machine (SSVM), we formulate a model that
combines different types of potential functions, including one that classifies
image regions using deep learning. Our main goal with this work is to show the
accuracy and efficiency improvements that these relatively new techniques can
provide for the segmentation of breast masses from mammograms. We also propose
an easily reproducible quantitative analysis to as- sess the performance of
breast mass segmentation methodologies based on widely accepted accuracy and
running time measurements on public datasets, which will facilitate further
comparisons for this segmentation problem. In particular, we use two publicly
available datasets (DDSM-BCRP and INbreast) and propose the computa- tion of
the running time taken for the methodology to produce a mass segmentation given
an input image and the use of the Dice index to quantitatively measure the
segmentation accuracy. For both databases, we show that our proposed
methodology produces competitive results in terms of accuracy and running time.
| [
{
"version": "v1",
"created": "Mon, 27 Oct 2014 22:44:26 GMT"
},
{
"version": "v2",
"created": "Fri, 5 Dec 2014 01:21:39 GMT"
}
] | 2014-12-08T00:00:00 | [
[
"Dhungel",
"Neeraj",
""
],
[
"Carneiro",
"Gustavo",
""
],
[
"Bradley",
"Andrew P.",
""
]
] | TITLE: Deep Structured learning for mass segmentation from Mammograms
ABSTRACT: In this paper, we present a novel method for the segmentation of breast
masses from mammograms exploring structured and deep learning. Specifically,
using structured support vector machine (SSVM), we formulate a model that
combines different types of potential functions, including one that classifies
image regions using deep learning. Our main goal with this work is to show the
accuracy and efficiency improvements that these relatively new techniques can
provide for the segmentation of breast masses from mammograms. We also propose
an easily reproducible quantitative analysis to as- sess the performance of
breast mass segmentation methodologies based on widely accepted accuracy and
running time measurements on public datasets, which will facilitate further
comparisons for this segmentation problem. In particular, we use two publicly
available datasets (DDSM-BCRP and INbreast) and propose the computa- tion of
the running time taken for the methodology to produce a mass segmentation given
an input image and the use of the Dice index to quantitatively measure the
segmentation accuracy. For both databases, we show that our proposed
methodology produces competitive results in terms of accuracy and running time.
|
1412.1842 | Max Jaderberg | Max Jaderberg, Karen Simonyan, Andrea Vedaldi, Andrew Zisserman | Reading Text in the Wild with Convolutional Neural Networks | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we present an end-to-end system for text spotting -- localising
and recognising text in natural scene images -- and text based image retrieval.
This system is based on a region proposal mechanism for detection and deep
convolutional neural networks for recognition. Our pipeline uses a novel
combination of complementary proposal generation techniques to ensure high
recall, and a fast subsequent filtering stage for improving precision. For the
recognition and ranking of proposals, we train very large convolutional neural
networks to perform word recognition on the whole proposal region at the same
time, departing from the character classifier based systems of the past. These
networks are trained solely on data produced by a synthetic text generation
engine, requiring no human labelled data.
Analysing the stages of our pipeline, we show state-of-the-art performance
throughout. We perform rigorous experiments across a number of standard
end-to-end text spotting benchmarks and text-based image retrieval datasets,
showing a large improvement over all previous methods. Finally, we demonstrate
a real-world application of our text spotting system to allow thousands of
hours of news footage to be instantly searchable via a text query.
| [
{
"version": "v1",
"created": "Thu, 4 Dec 2014 21:14:59 GMT"
}
] | 2014-12-08T00:00:00 | [
[
"Jaderberg",
"Max",
""
],
[
"Simonyan",
"Karen",
""
],
[
"Vedaldi",
"Andrea",
""
],
[
"Zisserman",
"Andrew",
""
]
] | TITLE: Reading Text in the Wild with Convolutional Neural Networks
ABSTRACT: In this work we present an end-to-end system for text spotting -- localising
and recognising text in natural scene images -- and text based image retrieval.
This system is based on a region proposal mechanism for detection and deep
convolutional neural networks for recognition. Our pipeline uses a novel
combination of complementary proposal generation techniques to ensure high
recall, and a fast subsequent filtering stage for improving precision. For the
recognition and ranking of proposals, we train very large convolutional neural
networks to perform word recognition on the whole proposal region at the same
time, departing from the character classifier based systems of the past. These
networks are trained solely on data produced by a synthetic text generation
engine, requiring no human labelled data.
Analysing the stages of our pipeline, we show state-of-the-art performance
throughout. We perform rigorous experiments across a number of standard
end-to-end text spotting benchmarks and text-based image retrieval datasets,
showing a large improvement over all previous methods. Finally, we demonstrate
a real-world application of our text spotting system to allow thousands of
hours of news footage to be instantly searchable via a text query.
|
1412.1888 | Rafi Muhammad | Muhammad Rafi, Farnaz Amin, Mohammad Shahid Shaikh | Document clustering using graph based document representation with
constraints | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Document clustering is an unsupervised approach in which a large collection
of documents (corpus) is subdivided into smaller, meaningful, identifiable, and
verifiable sub-groups (clusters). Meaningful representation of documents and
implicitly identifying the patterns, on which this separation is performed, is
the challenging part of document clustering. We have proposed a document
clustering technique using graph based document representation with
constraints. A graph data structure can easily capture the non-linear
relationships of nodes, document contains various feature terms that can be
non-linearly connected hence a graph can easily represents this information.
Constrains, are explicit conditions for document clustering where background
knowledge is use to set the direction for Linking or Not-Linking a set of
documents for a target clusters, thus guiding the clustering process. We deemed
clustering is an ill-define problem, there can be many clustering results.
Background knowledge can be used to drive the clustering algorithm in the right
direction. We have proposed three different types of constraints, Instance
level, corpus level and cluster level constraints. A new algorithm Constrained
HAC is also proposed which will incorporate Instance level constraints as prior
knowledge; it will guide the clustering process leading to better results.
Extensive set of experiments have been performed on both synthetic and standard
document clustering datasets, results are compared on standard clustering
measures like: purity, entropy and F-measure. Results clearly establish that
our proposed approach leads to improvement in cluster quality.
| [
{
"version": "v1",
"created": "Fri, 5 Dec 2014 03:47:34 GMT"
}
] | 2014-12-08T00:00:00 | [
[
"Rafi",
"Muhammad",
""
],
[
"Amin",
"Farnaz",
""
],
[
"Shaikh",
"Mohammad Shahid",
""
]
] | TITLE: Document clustering using graph based document representation with
constraints
ABSTRACT: Document clustering is an unsupervised approach in which a large collection
of documents (corpus) is subdivided into smaller, meaningful, identifiable, and
verifiable sub-groups (clusters). Meaningful representation of documents and
implicitly identifying the patterns, on which this separation is performed, is
the challenging part of document clustering. We have proposed a document
clustering technique using graph based document representation with
constraints. A graph data structure can easily capture the non-linear
relationships of nodes, document contains various feature terms that can be
non-linearly connected hence a graph can easily represents this information.
Constrains, are explicit conditions for document clustering where background
knowledge is use to set the direction for Linking or Not-Linking a set of
documents for a target clusters, thus guiding the clustering process. We deemed
clustering is an ill-define problem, there can be many clustering results.
Background knowledge can be used to drive the clustering algorithm in the right
direction. We have proposed three different types of constraints, Instance
level, corpus level and cluster level constraints. A new algorithm Constrained
HAC is also proposed which will incorporate Instance level constraints as prior
knowledge; it will guide the clustering process leading to better results.
Extensive set of experiments have been performed on both synthetic and standard
document clustering datasets, results are compared on standard clustering
measures like: purity, entropy and F-measure. Results clearly establish that
our proposed approach leads to improvement in cluster quality.
|
1412.1908 | Rui Zhao | Rui Zhao, Wanli Ouyang, Xiaogang Wang | Person Re-identification by Saliency Learning | This manuscript has 14 pages with 25 figures, and a preliminary
version was published in ICCV 2013 | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/3.0/ | Human eyes can recognize person identities based on small salient regions,
i.e. human saliency is distinctive and reliable in pedestrian matching across
disjoint camera views. However, such valuable information is often hidden when
computing similarities of pedestrian images with existing approaches. Inspired
by our user study result of human perception on human saliency, we propose a
novel perspective for person re-identification based on learning human saliency
and matching saliency distribution. The proposed saliency learning and matching
framework consists of four steps: (1) To handle misalignment caused by drastic
viewpoint change and pose variations, we apply adjacency constrained patch
matching to build dense correspondence between image pairs. (2) We propose two
alternative methods, i.e. K-Nearest Neighbors and One-class SVM, to estimate a
saliency score for each image patch, through which distinctive features stand
out without using identity labels in the training procedure. (3) saliency
matching is proposed based on patch matching. Matching patches with
inconsistent saliency brings penalty, and images of the same identity are
recognized by minimizing the saliency matching cost. (4) Furthermore, saliency
matching is tightly integrated with patch matching in a unified structural
RankSVM learning framework. The effectiveness of our approach is validated on
the VIPeR dataset and the CUHK01 dataset. Our approach outperforms the
state-of-the-art person re-identification methods on both datasets.
| [
{
"version": "v1",
"created": "Fri, 5 Dec 2014 07:33:48 GMT"
}
] | 2014-12-08T00:00:00 | [
[
"Zhao",
"Rui",
""
],
[
"Ouyang",
"Wanli",
""
],
[
"Wang",
"Xiaogang",
""
]
] | TITLE: Person Re-identification by Saliency Learning
ABSTRACT: Human eyes can recognize person identities based on small salient regions,
i.e. human saliency is distinctive and reliable in pedestrian matching across
disjoint camera views. However, such valuable information is often hidden when
computing similarities of pedestrian images with existing approaches. Inspired
by our user study result of human perception on human saliency, we propose a
novel perspective for person re-identification based on learning human saliency
and matching saliency distribution. The proposed saliency learning and matching
framework consists of four steps: (1) To handle misalignment caused by drastic
viewpoint change and pose variations, we apply adjacency constrained patch
matching to build dense correspondence between image pairs. (2) We propose two
alternative methods, i.e. K-Nearest Neighbors and One-class SVM, to estimate a
saliency score for each image patch, through which distinctive features stand
out without using identity labels in the training procedure. (3) saliency
matching is proposed based on patch matching. Matching patches with
inconsistent saliency brings penalty, and images of the same identity are
recognized by minimizing the saliency matching cost. (4) Furthermore, saliency
matching is tightly integrated with patch matching in a unified structural
RankSVM learning framework. The effectiveness of our approach is validated on
the VIPeR dataset and the CUHK01 dataset. Our approach outperforms the
state-of-the-art person re-identification methods on both datasets.
|
1412.1947 | Aditya AV Sastry Mr. | Aditya AV Sastry and Kalyan Netti | A parallel sampling based clustering | null | null | null | null | cs.LG | http://creativecommons.org/licenses/publicdomain/ | The problem of automatically clustering data is an age old problem. People
have created numerous algorithms to tackle this problem. The execution time of
any of this algorithm grows with the number of input points and the number of
cluster centers required. To reduce the number of input points we could average
the points locally and use the means or the local centers as the input for
clustering. However since the required number of local centers is very high,
running the clustering algorithm on the entire dataset to obtain these
representational points is very time consuming. To remedy this problem, in this
paper we are proposing two subclustering schemes where by we subdivide the
dataset into smaller sets and run the clustering algorithm on the smaller
datasets to obtain the required number of datapoints to run our clustering
algorithm with. As we are subdividing the given dataset, we could run
clustering algorithm on each smaller piece of the dataset in parallel. We found
that both parallel and serial execution of this method to be much faster than
the original clustering algorithm and error in running the clustering algorithm
on a reduced set to be very less.
| [
{
"version": "v1",
"created": "Fri, 5 Dec 2014 10:50:31 GMT"
}
] | 2014-12-08T00:00:00 | [
[
"Sastry",
"Aditya AV",
""
],
[
"Netti",
"Kalyan",
""
]
] | TITLE: A parallel sampling based clustering
ABSTRACT: The problem of automatically clustering data is an age old problem. People
have created numerous algorithms to tackle this problem. The execution time of
any of this algorithm grows with the number of input points and the number of
cluster centers required. To reduce the number of input points we could average
the points locally and use the means or the local centers as the input for
clustering. However since the required number of local centers is very high,
running the clustering algorithm on the entire dataset to obtain these
representational points is very time consuming. To remedy this problem, in this
paper we are proposing two subclustering schemes where by we subdivide the
dataset into smaller sets and run the clustering algorithm on the smaller
datasets to obtain the required number of datapoints to run our clustering
algorithm with. As we are subdividing the given dataset, we could run
clustering algorithm on each smaller piece of the dataset in parallel. We found
that both parallel and serial execution of this method to be much faster than
the original clustering algorithm and error in running the clustering algorithm
on a reduced set to be very less.
|
1306.3534 | Ashish Vulimiri | Ashish Vulimiri, P. Brighten Godfrey, Sri Varsha Gorge, Zitian Liu and
Scott Shenker | A cost-benefit analysis of low latency via added utilization | null | null | null | null | cs.NI cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Several recently proposed techniques achieve latency reduction by trading it
off for some amount of additional bandwidth usage. But how would one quantify
whether the tradeoff is actually beneficial in a given system? We develop an
economic cost vs. benefit analysis for answering this question. We use the
analysis to derive a benchmark for wide-area client-server applications, and
demonstrate how it can be applied to reason about a particular latency saving
technique --- redundant DNS requests.
| [
{
"version": "v1",
"created": "Sat, 15 Jun 2013 00:31:56 GMT"
},
{
"version": "v2",
"created": "Mon, 10 Nov 2014 16:56:41 GMT"
},
{
"version": "v3",
"created": "Fri, 14 Nov 2014 12:38:20 GMT"
},
{
"version": "v4",
"created": "Thu, 4 Dec 2014 20:53:48 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Vulimiri",
"Ashish",
""
],
[
"Godfrey",
"P. Brighten",
""
],
[
"Gorge",
"Sri Varsha",
""
],
[
"Liu",
"Zitian",
""
],
[
"Shenker",
"Scott",
""
]
] | TITLE: A cost-benefit analysis of low latency via added utilization
ABSTRACT: Several recently proposed techniques achieve latency reduction by trading it
off for some amount of additional bandwidth usage. But how would one quantify
whether the tradeoff is actually beneficial in a given system? We develop an
economic cost vs. benefit analysis for answering this question. We use the
analysis to derive a benchmark for wide-area client-server applications, and
demonstrate how it can be applied to reason about a particular latency saving
technique --- redundant DNS requests.
|
1404.1602 | James Pallister | James Pallister, Kerstin Eder, Simon J. Hollis, Jeremy Bennett | A high-level model of embedded flash energy consumption | null | null | 10.1145/2656106.2656108 | null | cs.AR | http://creativecommons.org/licenses/by-nc-sa/3.0/ | The alignment of code in the flash memory of deeply embedded SoCs can have a
large impact on the total energy consumption of a computation. We investigate
the effect of code alignment in six SoCs and find that a large proportion of
this energy (up to 15% of total SoC energy consumption) can be saved by changes
to the alignment.
A flexible model is created to predict the read-access energy consumption of
flash memory on deeply embedded SoCs, where code is executed in place. This
model uses the instruction level memory accesses performed by the processor to
calculate the flash energy consumption of a sequence of instructions. We derive
the model parameters for five SoCs and validate them. The error is as low as
5%, with a 11% average normalized RMS deviation overall.
The scope for using this model to optimize code alignment is explored across
a range of benchmarks and SoCs. Analysis shows that over 30% of loops can be
better aligned. This can significantly reduce energy while increasing code size
by less than 4%. We conclude that this effect has potential as an effective
optimization, saving significant energy in deeply embedded SoCs.
| [
{
"version": "v1",
"created": "Sun, 6 Apr 2014 17:34:23 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Apr 2014 08:53:35 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Pallister",
"James",
""
],
[
"Eder",
"Kerstin",
""
],
[
"Hollis",
"Simon J.",
""
],
[
"Bennett",
"Jeremy",
""
]
] | TITLE: A high-level model of embedded flash energy consumption
ABSTRACT: The alignment of code in the flash memory of deeply embedded SoCs can have a
large impact on the total energy consumption of a computation. We investigate
the effect of code alignment in six SoCs and find that a large proportion of
this energy (up to 15% of total SoC energy consumption) can be saved by changes
to the alignment.
A flexible model is created to predict the read-access energy consumption of
flash memory on deeply embedded SoCs, where code is executed in place. This
model uses the instruction level memory accesses performed by the processor to
calculate the flash energy consumption of a sequence of instructions. We derive
the model parameters for five SoCs and validate them. The error is as low as
5%, with a 11% average normalized RMS deviation overall.
The scope for using this model to optimize code alignment is explored across
a range of benchmarks and SoCs. Analysis shows that over 30% of loops can be
better aligned. This can significantly reduce energy while increasing code size
by less than 4%. We conclude that this effect has potential as an effective
optimization, saving significant energy in deeply embedded SoCs.
|
1409.1801 | Stephen Plaza | Stephen M. Plaza, Toufiq Parag, Gary B. Huang, Donald J. Olbris,
Mathew A. Saunders, Patricia K. Rivlin | Annotating Synapses in Large EM Datasets | null | null | null | null | q-bio.QM cs.CV q-bio.NC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reconstructing neuronal circuits at the level of synapses is a central
problem in neuroscience and becoming a focus of the emerging field of
connectomics. To date, electron microscopy (EM) is the most proven technique
for identifying and quantifying synaptic connections. As advances in EM make
acquiring larger datasets possible, subsequent manual synapse identification
({\em i.e.}, proofreading) for deciphering a connectome becomes a major time
bottleneck. Here we introduce a large-scale, high-throughput, and
semi-automated methodology to efficiently identify synapses. We successfully
applied our methodology to the Drosophila medulla optic lobe, annotating many
more synapses than previous connectome efforts. Our approaches are extensible
and will make the often complicated process of synapse identification
accessible to a wider-community of potential proofreaders.
| [
{
"version": "v1",
"created": "Fri, 5 Sep 2014 13:52:47 GMT"
},
{
"version": "v2",
"created": "Thu, 4 Dec 2014 16:18:01 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Plaza",
"Stephen M.",
""
],
[
"Parag",
"Toufiq",
""
],
[
"Huang",
"Gary B.",
""
],
[
"Olbris",
"Donald J.",
""
],
[
"Saunders",
"Mathew A.",
""
],
[
"Rivlin",
"Patricia K.",
""
]
] | TITLE: Annotating Synapses in Large EM Datasets
ABSTRACT: Reconstructing neuronal circuits at the level of synapses is a central
problem in neuroscience and becoming a focus of the emerging field of
connectomics. To date, electron microscopy (EM) is the most proven technique
for identifying and quantifying synaptic connections. As advances in EM make
acquiring larger datasets possible, subsequent manual synapse identification
({\em i.e.}, proofreading) for deciphering a connectome becomes a major time
bottleneck. Here we introduce a large-scale, high-throughput, and
semi-automated methodology to efficiently identify synapses. We successfully
applied our methodology to the Drosophila medulla optic lobe, annotating many
more synapses than previous connectome efforts. Our approaches are extensible
and will make the often complicated process of synapse identification
accessible to a wider-community of potential proofreaders.
|
1412.1602 | Jan Chorowski | Jan Chorowski, Dzmitry Bahdanau, Kyunghyun Cho, Yoshua Bengio | End-to-end Continuous Speech Recognition using Attention-based Recurrent
NN: First Results | As accepted to: Deep Learning and Representation Learning Workshop,
NIPS 2014 | null | null | null | cs.NE cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We replace the Hidden Markov Model (HMM) which is traditionally used in in
continuous speech recognition with a bi-directional recurrent neural network
encoder coupled to a recurrent neural network decoder that directly emits a
stream of phonemes. The alignment between the input and output sequences is
established using an attention mechanism: the decoder emits each symbol based
on a context created with a subset of input symbols elected by the attention
mechanism. We report initial results demonstrating that this new approach
achieves phoneme error rates that are comparable to the state-of-the-art
HMM-based decoders, on the TIMIT dataset.
| [
{
"version": "v1",
"created": "Thu, 4 Dec 2014 10:00:19 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Chorowski",
"Jan",
""
],
[
"Bahdanau",
"Dzmitry",
""
],
[
"Cho",
"Kyunghyun",
""
],
[
"Bengio",
"Yoshua",
""
]
] | TITLE: End-to-end Continuous Speech Recognition using Attention-based Recurrent
NN: First Results
ABSTRACT: We replace the Hidden Markov Model (HMM) which is traditionally used in in
continuous speech recognition with a bi-directional recurrent neural network
encoder coupled to a recurrent neural network decoder that directly emits a
stream of phonemes. The alignment between the input and output sequences is
established using an attention mechanism: the decoder emits each symbol based
on a context created with a subset of input symbols elected by the attention
mechanism. We report initial results demonstrating that this new approach
achieves phoneme error rates that are comparable to the state-of-the-art
HMM-based decoders, on the TIMIT dataset.
|
1412.1632 | Lei Yu | Lei Yu, Karl Moritz Hermann, Phil Blunsom and Stephen Pulman | Deep Learning for Answer Sentence Selection | 9 pages, accepted by NIPS deep learning workshop | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Answer sentence selection is the task of identifying sentences that contain
the answer to a given question. This is an important problem in its own right
as well as in the larger context of open domain question answering. We propose
a novel approach to solving this task via means of distributed representations,
and learn to match questions with answers by considering their semantic
encoding. This contrasts prior work on this task, which typically relies on
classifiers with large numbers of hand-crafted syntactic and semantic features
and various external resources. Our approach does not require any feature
engineering nor does it involve specialist linguistic data, making this model
easily applicable to a wide range of domains and languages. Experimental
results on a standard benchmark dataset from TREC demonstrate that---despite
its simplicity---our model matches state of the art performance on the answer
sentence selection task.
| [
{
"version": "v1",
"created": "Thu, 4 Dec 2014 11:53:02 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Yu",
"Lei",
""
],
[
"Hermann",
"Karl Moritz",
""
],
[
"Blunsom",
"Phil",
""
],
[
"Pulman",
"Stephen",
""
]
] | TITLE: Deep Learning for Answer Sentence Selection
ABSTRACT: Answer sentence selection is the task of identifying sentences that contain
the answer to a given question. This is an important problem in its own right
as well as in the larger context of open domain question answering. We propose
a novel approach to solving this task via means of distributed representations,
and learn to match questions with answers by considering their semantic
encoding. This contrasts prior work on this task, which typically relies on
classifiers with large numbers of hand-crafted syntactic and semantic features
and various external resources. Our approach does not require any feature
engineering nor does it involve specialist linguistic data, making this model
easily applicable to a wide range of domains and languages. Experimental
results on a standard benchmark dataset from TREC demonstrate that---despite
its simplicity---our model matches state of the art performance on the answer
sentence selection task.
|
1412.1710 | Kaiming He | Kaiming He, Jian Sun | Convolutional Neural Networks at Constrained Time Cost | 8-page technical report | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Though recent advanced convolutional neural networks (CNNs) have been
improving the image recognition accuracy, the models are getting more complex
and time-consuming. For real-world applications in industrial and commercial
scenarios, engineers and developers are often faced with the requirement of
constrained time budget. In this paper, we investigate the accuracy of CNNs
under constrained time cost. Under this constraint, the designs of the network
architectures should exhibit as trade-offs among the factors like depth,
numbers of filters, filter sizes, etc. With a series of controlled comparisons,
we progressively modify a baseline model while preserving its time complexity.
This is also helpful for understanding the importance of the factors in network
designs. We present an architecture that achieves very competitive accuracy in
the ImageNet dataset (11.8% top-5 error, 10-view test), yet is 20% faster than
"AlexNet" (16.0% top-5 error, 10-view test).
| [
{
"version": "v1",
"created": "Thu, 4 Dec 2014 16:00:47 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"He",
"Kaiming",
""
],
[
"Sun",
"Jian",
""
]
] | TITLE: Convolutional Neural Networks at Constrained Time Cost
ABSTRACT: Though recent advanced convolutional neural networks (CNNs) have been
improving the image recognition accuracy, the models are getting more complex
and time-consuming. For real-world applications in industrial and commercial
scenarios, engineers and developers are often faced with the requirement of
constrained time budget. In this paper, we investigate the accuracy of CNNs
under constrained time cost. Under this constraint, the designs of the network
architectures should exhibit as trade-offs among the factors like depth,
numbers of filters, filter sizes, etc. With a series of controlled comparisons,
we progressively modify a baseline model while preserving its time complexity.
This is also helpful for understanding the importance of the factors in network
designs. We present an architecture that achieves very competitive accuracy in
the ImageNet dataset (11.8% top-5 error, 10-view test), yet is 20% faster than
"AlexNet" (16.0% top-5 error, 10-view test).
|
1412.1717 | Dhagash Mehta | Joseph Cleveland, Jeffrey Dzugan, Jonathan D. Hauenstein, Ian Haywood,
Dhagash Mehta, Anthony Morse, Leonardo Robol and Taylor Schlenk | Certified counting of roots of random univariate polynomials | 28 pages, 16 figures | null | null | null | math.NA cs.NA hep-th nlin.CD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A challenging problem in computational mathematics is to compute roots of a
high-degree univariate random polynomial. We combine an efficient
multiprecision implementation for solving high-degree random polynomials with
two certification methods, namely Smale's $\alpha$-theory and one based on
Gerschgorin's theorem, for showing that a given numerical approximation is in
the quadratic convergence region of Newton's method of some exact solution.
With this combination, we can certifiably count the number of real roots of
random polynomials. We quantify the difference between the two certification
procedures and list the salient features of both of them. After benchmarking on
random polynomials where the coefficients are drawn from the Gaussian
distribution, we obtain novel experimental results for the Cauchy distribution
case.
| [
{
"version": "v1",
"created": "Thu, 4 Dec 2014 16:20:06 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Cleveland",
"Joseph",
""
],
[
"Dzugan",
"Jeffrey",
""
],
[
"Hauenstein",
"Jonathan D.",
""
],
[
"Haywood",
"Ian",
""
],
[
"Mehta",
"Dhagash",
""
],
[
"Morse",
"Anthony",
""
],
[
"Robol",
"Leonardo",
""
],
[
"Schlenk",
"Taylor",
""
]
] | TITLE: Certified counting of roots of random univariate polynomials
ABSTRACT: A challenging problem in computational mathematics is to compute roots of a
high-degree univariate random polynomial. We combine an efficient
multiprecision implementation for solving high-degree random polynomials with
two certification methods, namely Smale's $\alpha$-theory and one based on
Gerschgorin's theorem, for showing that a given numerical approximation is in
the quadratic convergence region of Newton's method of some exact solution.
With this combination, we can certifiably count the number of real roots of
random polynomials. We quantify the difference between the two certification
procedures and list the salient features of both of them. After benchmarking on
random polynomials where the coefficients are drawn from the Gaussian
distribution, we obtain novel experimental results for the Cauchy distribution
case.
|
1412.1788 | Francis Bach | Felipe Yanez (LIENS, INRIA Paris - Rocquencourt), Francis Bach (LIENS,
INRIA Paris - Rocquencourt) | Primal-Dual Algorithms for Non-negative Matrix Factorization with the
Kullback-Leibler Divergence | null | null | null | null | cs.LG math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Non-negative matrix factorization (NMF) approximates a given matrix as a
product of two non-negative matrices. Multiplicative algorithms deliver
reliable results, but they show slow convergence for high-dimensional data and
may be stuck away from local minima. Gradient descent methods have better
behavior, but only apply to smooth losses such as the least-squares loss. In
this article, we propose a first-order primal-dual algorithm for non-negative
decomposition problems (where one factor is fixed) with the KL divergence,
based on the Chambolle-Pock algorithm. All required computations may be
obtained in closed form and we provide an efficient heuristic way to select
step-sizes. By using alternating optimization, our algorithm readily extends to
NMF and, on synthetic examples, face recognition or music source separation
datasets, it is either faster than existing algorithms, or leads to improved
local optima, or both.
| [
{
"version": "v1",
"created": "Thu, 4 Dec 2014 19:56:02 GMT"
}
] | 2014-12-05T00:00:00 | [
[
"Yanez",
"Felipe",
"",
"LIENS, INRIA Paris - Rocquencourt"
],
[
"Bach",
"Francis",
"",
"LIENS,\n INRIA Paris - Rocquencourt"
]
] | TITLE: Primal-Dual Algorithms for Non-negative Matrix Factorization with the
Kullback-Leibler Divergence
ABSTRACT: Non-negative matrix factorization (NMF) approximates a given matrix as a
product of two non-negative matrices. Multiplicative algorithms deliver
reliable results, but they show slow convergence for high-dimensional data and
may be stuck away from local minima. Gradient descent methods have better
behavior, but only apply to smooth losses such as the least-squares loss. In
this article, we propose a first-order primal-dual algorithm for non-negative
decomposition problems (where one factor is fixed) with the KL divergence,
based on the Chambolle-Pock algorithm. All required computations may be
obtained in closed form and we provide an efficient heuristic way to select
step-sizes. By using alternating optimization, our algorithm readily extends to
NMF and, on synthetic examples, face recognition or music source separation
datasets, it is either faster than existing algorithms, or leads to improved
local optima, or both.
|
1309.5643 | Veronika Cheplygina | Veronika Cheplygina, David M. J. Tax, and Marco Loog | Multiple Instance Learning with Bag Dissimilarities | Pattern Recognition, in press | Pattern Recognition 48.1 (2015): 264-275 | 10.1016/j.patcog.2014.07.022 | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multiple instance learning (MIL) is concerned with learning from sets (bags)
of objects (instances), where the individual instance labels are ambiguous. In
this setting, supervised learning cannot be applied directly. Often,
specialized MIL methods learn by making additional assumptions about the
relationship of the bag labels and instance labels. Such assumptions may fit a
particular dataset, but do not generalize to the whole range of MIL problems.
Other MIL methods shift the focus of assumptions from the labels to the overall
(dis)similarity of bags, and therefore learn from bags directly. We propose to
represent each bag by a vector of its dissimilarities to other bags in the
training set, and treat these dissimilarities as a feature representation. We
show several alternatives to define a dissimilarity between bags and discuss
which definitions are more suitable for particular MIL problems. The
experimental results show that the proposed approach is computationally
inexpensive, yet very competitive with state-of-the-art algorithms on a wide
range of MIL datasets.
| [
{
"version": "v1",
"created": "Sun, 22 Sep 2013 20:24:50 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Feb 2014 13:13:11 GMT"
},
{
"version": "v3",
"created": "Tue, 12 Aug 2014 09:04:32 GMT"
}
] | 2014-12-04T00:00:00 | [
[
"Cheplygina",
"Veronika",
""
],
[
"Tax",
"David M. J.",
""
],
[
"Loog",
"Marco",
""
]
] | TITLE: Multiple Instance Learning with Bag Dissimilarities
ABSTRACT: Multiple instance learning (MIL) is concerned with learning from sets (bags)
of objects (instances), where the individual instance labels are ambiguous. In
this setting, supervised learning cannot be applied directly. Often,
specialized MIL methods learn by making additional assumptions about the
relationship of the bag labels and instance labels. Such assumptions may fit a
particular dataset, but do not generalize to the whole range of MIL problems.
Other MIL methods shift the focus of assumptions from the labels to the overall
(dis)similarity of bags, and therefore learn from bags directly. We propose to
represent each bag by a vector of its dissimilarities to other bags in the
training set, and treat these dissimilarities as a feature representation. We
show several alternatives to define a dissimilarity between bags and discuss
which definitions are more suitable for particular MIL problems. The
experimental results show that the proposed approach is computationally
inexpensive, yet very competitive with state-of-the-art algorithms on a wide
range of MIL datasets.
|
1406.2348 | Szymon Grabowski | Szymon Grabowski, Marcin Raniszewski | Sampling the suffix array with minimizers | One new SamSAMi variant; extended experimental results | null | null | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sampling (evenly) the suffixes from the suffix array is an old idea trading
the pattern search time for reduced index space. A few years ago Claude et al.
showed an alphabet sampling scheme allowing for more efficient pattern searches
compared to the sparse suffix array, for long enough patterns. A drawback of
their approach is the requirement that sought patterns need to contain at least
one character from the chosen subalphabet. In this work we propose an
alternative suffix sampling approach with only a minimum pattern length as a
requirement, which seems more convenient in practice. Experiments show that our
algorithm achieves competitive time-space tradeoffs on most standard benchmark
data.
| [
{
"version": "v1",
"created": "Mon, 9 Jun 2014 20:36:01 GMT"
},
{
"version": "v2",
"created": "Wed, 3 Dec 2014 19:39:41 GMT"
}
] | 2014-12-04T00:00:00 | [
[
"Grabowski",
"Szymon",
""
],
[
"Raniszewski",
"Marcin",
""
]
] | TITLE: Sampling the suffix array with minimizers
ABSTRACT: Sampling (evenly) the suffixes from the suffix array is an old idea trading
the pattern search time for reduced index space. A few years ago Claude et al.
showed an alphabet sampling scheme allowing for more efficient pattern searches
compared to the sparse suffix array, for long enough patterns. A drawback of
their approach is the requirement that sought patterns need to contain at least
one character from the chosen subalphabet. In this work we propose an
alternative suffix sampling approach with only a minimum pattern length as a
requirement, which seems more convenient in practice. Experiments show that our
algorithm achieves competitive time-space tradeoffs on most standard benchmark
data.
|
1410.4345 | Joel Akeret | Joel Akeret, Lukas Gamper, Adam Amara, Alexandre Refregier (ETH
Zurich) | HOPE: A Python Just-In-Time compiler for astrophysical computations | Accepted for publication in Astronomy and Computing. 14 pages, 1
figure. The code is available at http://hope.phys.ethz.ch | null | null | null | astro-ph.IM cs.MS cs.PL physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Python programming language is becoming increasingly popular for
scientific applications due to its simplicity, versatility, and the broad range
of its libraries. A drawback of this dynamic language, however, is its low
runtime performance which limits its applicability for large simulations and
for the analysis of large data sets, as is common in astrophysics and
cosmology. While various frameworks have been developed to address this
limitation, most focus on covering the complete language set, and either force
the user to alter the code or are not able to reach the full speed of an
optimised native compiled language. In order to combine the ease of Python and
the speed of C++, we developed HOPE, a specialised Python just-in-time (JIT)
compiler designed for numerical astrophysical applications. HOPE focuses on a
subset of the language and is able to translate Python code into C++ while
performing numerical optimisation on mathematical expressions at runtime. To
enable the JIT compilation, the user only needs to add a decorator to the
function definition. We assess the performance of HOPE by performing a series
of benchmarks and compare its execution speed with that of plain Python, C++
and the other existing frameworks. We find that HOPE improves the performance
compared to plain Python by a factor of 2 to 120, achieves speeds comparable to
that of C++, and often exceeds the speed of the existing solutions. We discuss
the differences between HOPE and the other frameworks, as well as future
extensions of its capabilities. The fully documented HOPE package is available
at http://hope.phys.ethz.ch and is published under the GPLv3 license on PyPI
and GitHub.
| [
{
"version": "v1",
"created": "Thu, 16 Oct 2014 09:28:18 GMT"
},
{
"version": "v2",
"created": "Wed, 3 Dec 2014 15:08:33 GMT"
}
] | 2014-12-04T00:00:00 | [
[
"Akeret",
"Joel",
"",
"ETH\n Zurich"
],
[
"Gamper",
"Lukas",
"",
"ETH\n Zurich"
],
[
"Amara",
"Adam",
"",
"ETH\n Zurich"
],
[
"Refregier",
"Alexandre",
"",
"ETH\n Zurich"
]
] | TITLE: HOPE: A Python Just-In-Time compiler for astrophysical computations
ABSTRACT: The Python programming language is becoming increasingly popular for
scientific applications due to its simplicity, versatility, and the broad range
of its libraries. A drawback of this dynamic language, however, is its low
runtime performance which limits its applicability for large simulations and
for the analysis of large data sets, as is common in astrophysics and
cosmology. While various frameworks have been developed to address this
limitation, most focus on covering the complete language set, and either force
the user to alter the code or are not able to reach the full speed of an
optimised native compiled language. In order to combine the ease of Python and
the speed of C++, we developed HOPE, a specialised Python just-in-time (JIT)
compiler designed for numerical astrophysical applications. HOPE focuses on a
subset of the language and is able to translate Python code into C++ while
performing numerical optimisation on mathematical expressions at runtime. To
enable the JIT compilation, the user only needs to add a decorator to the
function definition. We assess the performance of HOPE by performing a series
of benchmarks and compare its execution speed with that of plain Python, C++
and the other existing frameworks. We find that HOPE improves the performance
compared to plain Python by a factor of 2 to 120, achieves speeds comparable to
that of C++, and often exceeds the speed of the existing solutions. We discuss
the differences between HOPE and the other frameworks, as well as future
extensions of its capabilities. The fully documented HOPE package is available
at http://hope.phys.ethz.ch and is published under the GPLv3 license on PyPI
and GitHub.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.