id
stringlengths 9
16
| submitter
stringlengths 3
64
⌀ | authors
stringlengths 5
6.63k
| title
stringlengths 7
245
| comments
stringlengths 1
482
⌀ | journal-ref
stringlengths 4
382
⌀ | doi
stringlengths 9
151
⌀ | report-no
stringclasses 984
values | categories
stringlengths 5
108
| license
stringclasses 9
values | abstract
stringlengths 83
3.41k
| versions
listlengths 1
20
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-04-11 00:00:00
| authors_parsed
sequencelengths 1
427
| prompt
stringlengths 166
3.49k
| label
stringclasses 2
values | prob
float64 0.5
0.98
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1504.04596 | Yadong Zhu | Yadong Zhu, Yanyan Lan, Jiafeng Guo, Xueqi Cheng | Structural Learning of Diverse Ranking | Discriminant Function, Diversity Feature, Learning Framework | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Relevance and diversity are both crucial criteria for an effective search
system. In this paper, we propose a unified learning framework for
simultaneously optimizing both relevance and diversity. Specifically, the
problem is formalized as a structural learning framework optimizing
Diversity-Correlated Evaluation Measures (DCEM), such as ERR-IA, a-NDCG and
NRBP. Within this framework, the discriminant function is defined to be a
bi-criteria objective maximizing the sum of the relevance scores and
dissimilarities (or diversity) among the documents. Relevance and diversity
features are utilized to define the relevance scores and dissimilarities,
respectively. Compared with traditional methods, the advantages of our approach
lie in that: (1) Directly optimizing DCEM as the loss function is more
fundamental for the task; (2) Our framework does not rely on explicit diversity
information such as subtopics, thus is more adaptive to real application; (3)
The representation of diversity as the feature-based scoring function is more
flexible to incorporate rich diversity-based features into the learning
framework. Extensive experiments on the public TREC datasets show that our
approach significantly outperforms state-of-the-art diversification approaches,
which validate the above advantages.
| [
{
"version": "v1",
"created": "Fri, 17 Apr 2015 18:23:16 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Apr 2015 09:59:00 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Zhu",
"Yadong",
""
],
[
"Lan",
"Yanyan",
""
],
[
"Guo",
"Jiafeng",
""
],
[
"Cheng",
"Xueqi",
""
]
] | TITLE: Structural Learning of Diverse Ranking
ABSTRACT: Relevance and diversity are both crucial criteria for an effective search
system. In this paper, we propose a unified learning framework for
simultaneously optimizing both relevance and diversity. Specifically, the
problem is formalized as a structural learning framework optimizing
Diversity-Correlated Evaluation Measures (DCEM), such as ERR-IA, a-NDCG and
NRBP. Within this framework, the discriminant function is defined to be a
bi-criteria objective maximizing the sum of the relevance scores and
dissimilarities (or diversity) among the documents. Relevance and diversity
features are utilized to define the relevance scores and dissimilarities,
respectively. Compared with traditional methods, the advantages of our approach
lie in that: (1) Directly optimizing DCEM as the loss function is more
fundamental for the task; (2) Our framework does not rely on explicit diversity
information such as subtopics, thus is more adaptive to real application; (3)
The representation of diversity as the feature-based scoring function is more
flexible to incorporate rich diversity-based features into the learning
framework. Extensive experiments on the public TREC datasets show that our
approach significantly outperforms state-of-the-art diversification approaches,
which validate the above advantages.
| no_new_dataset | 0.946794 |
1504.04646 | Kwetishe Danjuma | Kwetishe Joro Danjuma | Performance Evaluation of Machine Learning Algorithms in Post-operative
Life Expectancy in the Lung Cancer Patients | 11 pages, 3 figures, 2 tables, ISSN (Print): 1694-0814 | ISSN
(Online): 1694-0784 | IJCSI International Journal of Computer Science Issues, Volume 12,
Issue 2, March 2015 | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The nature of clinical data makes it difficult to quickly select, tune and
apply machine learning algorithms to clinical prognosis. As a result, a lot of
time is spent searching for the most appropriate machine learning algorithms
applicable in clinical prognosis that contains either binary-valued or
multi-valued attributes. The study set out to identify and evaluate the
performance of machine learning classification schemes applied in clinical
prognosis of post-operative life expectancy in the lung cancer patients.
Multilayer Perceptron, J48, and the Naive Bayes algorithms were used to train
and test models on Thoracic Surgery datasets obtained from the University of
California Irvine machine learning repository. Stratified 10-fold
cross-validation was used to evaluate baseline performance accuracy of the
classifiers. The comparative analysis shows that multilayer perceptron
performed best with classification accuracy of 82.3%, J48 came out second with
classification accuracy of 81.8%, and Naive Bayes came out the worst with
classification accuracy of 74.4%. The quality and outcome of the chosen machine
learning algorithms depends on the ingenuity of the clinical miner.
| [
{
"version": "v1",
"created": "Fri, 17 Apr 2015 22:05:34 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Danjuma",
"Kwetishe Joro",
""
]
] | TITLE: Performance Evaluation of Machine Learning Algorithms in Post-operative
Life Expectancy in the Lung Cancer Patients
ABSTRACT: The nature of clinical data makes it difficult to quickly select, tune and
apply machine learning algorithms to clinical prognosis. As a result, a lot of
time is spent searching for the most appropriate machine learning algorithms
applicable in clinical prognosis that contains either binary-valued or
multi-valued attributes. The study set out to identify and evaluate the
performance of machine learning classification schemes applied in clinical
prognosis of post-operative life expectancy in the lung cancer patients.
Multilayer Perceptron, J48, and the Naive Bayes algorithms were used to train
and test models on Thoracic Surgery datasets obtained from the University of
California Irvine machine learning repository. Stratified 10-fold
cross-validation was used to evaluate baseline performance accuracy of the
classifiers. The comparative analysis shows that multilayer perceptron
performed best with classification accuracy of 82.3%, J48 came out second with
classification accuracy of 81.8%, and Naive Bayes came out the worst with
classification accuracy of 74.4%. The quality and outcome of the chosen machine
learning algorithms depends on the ingenuity of the clinical miner.
| no_new_dataset | 0.946941 |
1504.04730 | Sourav Bhattacharya | Sourav Bhattacharya and Otto Huhta and N. Asokan | LookAhead: Augmenting Crowdsourced Website Reputation Systems With
Predictive Modeling | 12 pages | null | null | null | cs.CR cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Unsafe websites consist of malicious as well as inappropriate sites, such as
those hosting questionable or offensive content. Website reputation systems are
intended to help ordinary users steer away from these unsafe sites. However,
the process of assigning safety ratings for websites typically involves humans.
Consequently it is time consuming, costly and not scalable. This has resulted
in two major problems: (i) a significant proportion of the web space remains
unrated and (ii) there is an unacceptable time lag before new websites are
rated.
In this paper, we show that by leveraging structural and content-based
properties of websites, it is possible to reliably and efficiently predict
their safety ratings, thereby mitigating both problems. We demonstrate the
effectiveness of our approach using four datasets of up to 90,000 websites. We
use ratings from Web of Trust (WOT), a popular crowdsourced web reputation
system, as ground truth. We propose a novel ensemble classification technique
that makes opportunistic use of available structural and content properties of
webpages to predict their eventual ratings in two dimensions used by WOT:
trustworthiness and child safety. Ours is the first classification system to
predict such subjective ratings and the same approach works equally well in
identifying malicious websites. Across all datasets, our classification
performs well with average F$_1$-score in the 74--90\% range.
| [
{
"version": "v1",
"created": "Sat, 18 Apr 2015 15:13:13 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Bhattacharya",
"Sourav",
""
],
[
"Huhta",
"Otto",
""
],
[
"Asokan",
"N.",
""
]
] | TITLE: LookAhead: Augmenting Crowdsourced Website Reputation Systems With
Predictive Modeling
ABSTRACT: Unsafe websites consist of malicious as well as inappropriate sites, such as
those hosting questionable or offensive content. Website reputation systems are
intended to help ordinary users steer away from these unsafe sites. However,
the process of assigning safety ratings for websites typically involves humans.
Consequently it is time consuming, costly and not scalable. This has resulted
in two major problems: (i) a significant proportion of the web space remains
unrated and (ii) there is an unacceptable time lag before new websites are
rated.
In this paper, we show that by leveraging structural and content-based
properties of websites, it is possible to reliably and efficiently predict
their safety ratings, thereby mitigating both problems. We demonstrate the
effectiveness of our approach using four datasets of up to 90,000 websites. We
use ratings from Web of Trust (WOT), a popular crowdsourced web reputation
system, as ground truth. We propose a novel ensemble classification technique
that makes opportunistic use of available structural and content properties of
webpages to predict their eventual ratings in two dimensions used by WOT:
trustworthiness and child safety. Ours is the first classification system to
predict such subjective ratings and the same approach works equally well in
identifying malicious websites. Across all datasets, our classification
performs well with average F$_1$-score in the 74--90\% range.
| no_new_dataset | 0.949716 |
1504.04739 | Wojciech Czarnecki | Rafal Jozefowicz, Wojciech Marian Czarnecki | Fast optimization of Multithreshold Entropy Linear Classifier | Presented at Theoretical Foundations of Machine Learning 2015
(http://tfml.gmum.net), final version published in Schedae Informaticae
Journal | null | 10.4467/20838476SI.14.005.3022 | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multithreshold Entropy Linear Classifier (MELC) is a density based model
which searches for a linear projection maximizing the Cauchy-Schwarz Divergence
of dataset kernel density estimation. Despite its good empirical results, one
of its drawbacks is the optimization speed. In this paper we analyze how one
can speed it up through solving an approximate problem. We analyze two methods,
both similar to the approximate solutions of the Kernel Density Estimation
querying and provide adaptive schemes for selecting a crucial parameters based
on user-specified acceptable error. Furthermore we show how one can exploit
well known conjugate gradients and L-BFGS optimizers despite the fact that the
original optimization problem should be solved on the sphere. All above methods
and modifications are tested on 10 real life datasets from UCI repository to
confirm their practical usability.
| [
{
"version": "v1",
"created": "Sat, 18 Apr 2015 16:19:22 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Jozefowicz",
"Rafal",
""
],
[
"Czarnecki",
"Wojciech Marian",
""
]
] | TITLE: Fast optimization of Multithreshold Entropy Linear Classifier
ABSTRACT: Multithreshold Entropy Linear Classifier (MELC) is a density based model
which searches for a linear projection maximizing the Cauchy-Schwarz Divergence
of dataset kernel density estimation. Despite its good empirical results, one
of its drawbacks is the optimization speed. In this paper we analyze how one
can speed it up through solving an approximate problem. We analyze two methods,
both similar to the approximate solutions of the Kernel Density Estimation
querying and provide adaptive schemes for selecting a crucial parameters based
on user-specified acceptable error. Furthermore we show how one can exploit
well known conjugate gradients and L-BFGS optimizers despite the fact that the
original optimization problem should be solved on the sphere. All above methods
and modifications are tested on 10 real life datasets from UCI repository to
confirm their practical usability.
| no_new_dataset | 0.94743 |
1504.04740 | Wojciech Czarnecki | Wojciech Marian Czarnecki | On the consistency of Multithreshold Entropy Linear Classifier | Presented at Theoretical Foundations of Machine Learning 2015
(http://tfml.gmum.net), final version published in Schedae Informaticae
Journal | null | 10.4467/20838476SI.15.012.3034 | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multithreshold Entropy Linear Classifier (MELC) is a recent classifier idea
which employs information theoretic concept in order to create a multithreshold
maximum margin model. In this paper we analyze its consistency over
multithreshold linear models and show that its objective function upper bounds
the amount of misclassified points in a similar manner like hinge loss does in
support vector machines. For further confirmation we also conduct some
numerical experiments on five datasets.
| [
{
"version": "v1",
"created": "Sat, 18 Apr 2015 16:29:26 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Czarnecki",
"Wojciech Marian",
""
]
] | TITLE: On the consistency of Multithreshold Entropy Linear Classifier
ABSTRACT: Multithreshold Entropy Linear Classifier (MELC) is a recent classifier idea
which employs information theoretic concept in order to create a multithreshold
maximum margin model. In this paper we analyze its consistency over
multithreshold linear models and show that its objective function upper bounds
the amount of misclassified points in a similar manner like hinge loss does in
support vector machines. For further confirmation we also conduct some
numerical experiments on five datasets.
| no_new_dataset | 0.955693 |
1504.04871 | Sukrit Shankar | Sukrit Shankar, Vikas K. Garg, Roberto Cipolla | DEEP-CARVING: Discovering Visual Attributes by Carving Deep Neural Nets | 10 pages, 8 figures, CVPR 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most of the approaches for discovering visual attributes in images demand
significant supervision, which is cumbersome to obtain. In this paper, we aim
to discover visual attributes in a weakly supervised setting that is commonly
encountered with contemporary image search engines. Deep Convolutional Neural
Networks (CNNs) have enjoyed remarkable success in vision applications
recently. However, in a weakly supervised scenario, widely used CNN training
procedures do not learn a robust model for predicting multiple attribute labels
simultaneously. The primary reason is that the attributes highly co-occur
within the training data. To ameliorate this limitation, we propose
Deep-Carving, a novel training procedure with CNNs, that helps the net
efficiently carve itself for the task of multiple attribute prediction. During
training, the responses of the feature maps are exploited in an ingenious way
to provide the net with multiple pseudo-labels (for training images) for
subsequent iterations. The process is repeated periodically after a fixed
number of iterations, and enables the net carve itself iteratively for
efficiently disentangling features. Additionally, we contribute a
noun-adjective pairing inspired Natural Scenes Attributes Dataset to the
research community, CAMIT - NSAD, containing a number of co-occurring
attributes within a noun category. We describe, in detail, salient aspects of
this dataset. Our experiments on CAMIT-NSAD and the SUN Attributes Dataset,
with weak supervision, clearly demonstrate that the Deep-Carved CNNs
consistently achieve considerable improvement in the precision of attribute
prediction over popular baseline methods.
| [
{
"version": "v1",
"created": "Sun, 19 Apr 2015 18:56:52 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Shankar",
"Sukrit",
""
],
[
"Garg",
"Vikas K.",
""
],
[
"Cipolla",
"Roberto",
""
]
] | TITLE: DEEP-CARVING: Discovering Visual Attributes by Carving Deep Neural Nets
ABSTRACT: Most of the approaches for discovering visual attributes in images demand
significant supervision, which is cumbersome to obtain. In this paper, we aim
to discover visual attributes in a weakly supervised setting that is commonly
encountered with contemporary image search engines. Deep Convolutional Neural
Networks (CNNs) have enjoyed remarkable success in vision applications
recently. However, in a weakly supervised scenario, widely used CNN training
procedures do not learn a robust model for predicting multiple attribute labels
simultaneously. The primary reason is that the attributes highly co-occur
within the training data. To ameliorate this limitation, we propose
Deep-Carving, a novel training procedure with CNNs, that helps the net
efficiently carve itself for the task of multiple attribute prediction. During
training, the responses of the feature maps are exploited in an ingenious way
to provide the net with multiple pseudo-labels (for training images) for
subsequent iterations. The process is repeated periodically after a fixed
number of iterations, and enables the net carve itself iteratively for
efficiently disentangling features. Additionally, we contribute a
noun-adjective pairing inspired Natural Scenes Attributes Dataset to the
research community, CAMIT - NSAD, containing a number of co-occurring
attributes within a noun category. We describe, in detail, salient aspects of
this dataset. Our experiments on CAMIT-NSAD and the SUN Attributes Dataset,
with weak supervision, clearly demonstrate that the Deep-Carved CNNs
consistently achieve considerable improvement in the precision of attribute
prediction over popular baseline methods.
| no_new_dataset | 0.717631 |
1504.04923 | Chunhua Shen | Ruizhi Qiao, Lingqiao Liu, Chunhua Shen, and Anton von den Hengel | Learning discriminative trajectorylet detector sets for accurate
skeleton-based action recognition | 10 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The introduction of low-cost RGB-D sensors has promoted the research in
skeleton-based human action recognition. Devising a representation suitable for
characterising actions on the basis of noisy skeleton sequences remains a
challenge, however. We here provide two insights into this challenge. First, we
show that the discriminative information of a skeleton sequence usually resides
in a short temporal interval and we propose a simple-but-effective local
descriptor called trajectorylet to capture the static and kinematic information
within this interval. Second, we further propose to encode each trajectorylet
with a discriminative trajectorylet detector set which is selected from a large
number of candidate detectors trained through exemplar-SVMs. The action-level
representation is obtained by pooling trajectorylet encodings. Evaluating on
standard datasets acquired from the Kinect sensor, it is demonstrated that our
method obtains superior results over existing approaches under various
experimental setups.
| [
{
"version": "v1",
"created": "Mon, 20 Apr 2015 02:41:03 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Qiao",
"Ruizhi",
""
],
[
"Liu",
"Lingqiao",
""
],
[
"Shen",
"Chunhua",
""
],
[
"Hengel",
"Anton von den",
""
]
] | TITLE: Learning discriminative trajectorylet detector sets for accurate
skeleton-based action recognition
ABSTRACT: The introduction of low-cost RGB-D sensors has promoted the research in
skeleton-based human action recognition. Devising a representation suitable for
characterising actions on the basis of noisy skeleton sequences remains a
challenge, however. We here provide two insights into this challenge. First, we
show that the discriminative information of a skeleton sequence usually resides
in a short temporal interval and we propose a simple-but-effective local
descriptor called trajectorylet to capture the static and kinematic information
within this interval. Second, we further propose to encode each trajectorylet
with a discriminative trajectorylet detector set which is selected from a large
number of candidate detectors trained through exemplar-SVMs. The action-level
representation is obtained by pooling trajectorylet encodings. Evaluating on
standard datasets acquired from the Kinect sensor, it is demonstrated that our
method obtains superior results over existing approaches under various
experimental setups.
| no_new_dataset | 0.944228 |
1504.04945 | Yadong Zhu | Yadong Zhu, Yanyan Lan, Jiafeng Guo, Xueqi Cheng | Topic-focused Dynamic Information Filtering in Social Media | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the quick development of online social media such as twitter or sina
weibo in china, many users usually track hot topics to satisfy their desired
information need. For a hot topic, new opinions or ideas will be continuously
produced in the form of online data stream. In this scenario, how to
effectively filter and display information for a certain topic dynamically,
will be a critical problem. We call the problem as Topic-focused Dynamic
Information Filtering (denoted as TDIF for short) in social media. In this
paper, we start open discussions on such application problems. We first analyze
the properties of the TDIF problem, which usually contains several typical
requirements: relevance, diversity, recency and confidence. Recency means that
users want to follow the recent opinions or news. Additionally, the confidence
of information must be taken into consideration. How to balance these factors
properly in online data stream is very important and challenging. We propose a
dynamic preservation strategy on the basis of an existing feature-based utility
function, to solve the TDIF problem. Additionally, we propose new dynamic
diversity measures, to get a more reasonable evaluation for such application
problems. Extensive exploratory experiments have been conducted on TREC public
twitter dataset, and the experimental results validate the effectiveness of our
approach.
| [
{
"version": "v1",
"created": "Mon, 20 Apr 2015 06:12:06 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Zhu",
"Yadong",
""
],
[
"Lan",
"Yanyan",
""
],
[
"Guo",
"Jiafeng",
""
],
[
"Cheng",
"Xueqi",
""
]
] | TITLE: Topic-focused Dynamic Information Filtering in Social Media
ABSTRACT: With the quick development of online social media such as twitter or sina
weibo in china, many users usually track hot topics to satisfy their desired
information need. For a hot topic, new opinions or ideas will be continuously
produced in the form of online data stream. In this scenario, how to
effectively filter and display information for a certain topic dynamically,
will be a critical problem. We call the problem as Topic-focused Dynamic
Information Filtering (denoted as TDIF for short) in social media. In this
paper, we start open discussions on such application problems. We first analyze
the properties of the TDIF problem, which usually contains several typical
requirements: relevance, diversity, recency and confidence. Recency means that
users want to follow the recent opinions or news. Additionally, the confidence
of information must be taken into consideration. How to balance these factors
properly in online data stream is very important and challenging. We propose a
dynamic preservation strategy on the basis of an existing feature-based utility
function, to solve the TDIF problem. Additionally, we propose new dynamic
diversity measures, to get a more reasonable evaluation for such application
problems. Extensive exploratory experiments have been conducted on TREC public
twitter dataset, and the experimental results validate the effectiveness of our
approach.
| no_new_dataset | 0.952131 |
1504.05035 | Wangmeng Zuo | Xiaohe Wu, Wangmeng Zuo, Yuanyuan Zhu, Liang Lin | F-SVM: Combination of Feature Transformation and SVM Learning via Convex
Relaxation | 11 pages, 5 figures | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The generalization error bound of support vector machine (SVM) depends on the
ratio of radius and margin, while standard SVM only considers the maximization
of the margin but ignores the minimization of the radius. Several approaches
have been proposed to integrate radius and margin for joint learning of feature
transformation and SVM classifier. However, most of them either require the
form of the transformation matrix to be diagonal, or are non-convex and
computationally expensive. In this paper, we suggest a novel approximation for
the radius of minimum enclosing ball (MEB) in feature space, and then propose a
convex radius-margin based SVM model for joint learning of feature
transformation and SVM classifier, i.e., F-SVM. An alternating minimization
method is adopted to solve the F-SVM model, where the feature transformation is
updatedvia gradient descent and the classifier is updated by employing the
existing SVM solver. By incorporating with kernel principal component analysis,
F-SVM is further extended for joint learning of nonlinear transformation and
classifier. Experimental results on the UCI machine learning datasets and the
LFW face datasets show that F-SVM outperforms the standard SVM and the existing
radius-margin based SVMs, e.g., RMM, R-SVM+ and R-SVM+{\mu}.
| [
{
"version": "v1",
"created": "Mon, 20 Apr 2015 12:36:50 GMT"
}
] | 2015-04-21T00:00:00 | [
[
"Wu",
"Xiaohe",
""
],
[
"Zuo",
"Wangmeng",
""
],
[
"Zhu",
"Yuanyuan",
""
],
[
"Lin",
"Liang",
""
]
] | TITLE: F-SVM: Combination of Feature Transformation and SVM Learning via Convex
Relaxation
ABSTRACT: The generalization error bound of support vector machine (SVM) depends on the
ratio of radius and margin, while standard SVM only considers the maximization
of the margin but ignores the minimization of the radius. Several approaches
have been proposed to integrate radius and margin for joint learning of feature
transformation and SVM classifier. However, most of them either require the
form of the transformation matrix to be diagonal, or are non-convex and
computationally expensive. In this paper, we suggest a novel approximation for
the radius of minimum enclosing ball (MEB) in feature space, and then propose a
convex radius-margin based SVM model for joint learning of feature
transformation and SVM classifier, i.e., F-SVM. An alternating minimization
method is adopted to solve the F-SVM model, where the feature transformation is
updatedvia gradient descent and the classifier is updated by employing the
existing SVM solver. By incorporating with kernel principal component analysis,
F-SVM is further extended for joint learning of nonlinear transformation and
classifier. Experimental results on the UCI machine learning datasets and the
LFW face datasets show that F-SVM outperforms the standard SVM and the existing
radius-margin based SVMs, e.g., RMM, R-SVM+ and R-SVM+{\mu}.
| no_new_dataset | 0.954265 |
1305.1495 | Lorenzo Del Castello | Alessandro Attanasi, Andrea Cavagna, Lorenzo Del Castello, Irene
Giardina, Asja Jelic, Stefania Melillo, Leonardo Parisi, Fabio Pellacini,
Edward Shen, Edmondo Silvestri, Massimiliano Viale | GReTA - a novel Global and Recursive Tracking Algorithm in three
dimensions | 13 pages, 6 figures, 3 tables. Version 3 was slightly shortened, and
new comprative results on the public datasets (thermal infrared videos of
flying bats) by Z. Wu and coworkers (2014) were included. in A. Attanasi et
al., "GReTA - A Novel Global and Recursive Tracking Algorithm in Three
Dimensions", IEEE Trans. Pattern Anal. Mach. Intell., vol.37 (2015) | null | 10.1109/TPAMI.2015.2414427 | null | q-bio.QM cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Tracking multiple moving targets allows quantitative measure of the dynamic
behavior in systems as diverse as animal groups in biology, turbulence in fluid
dynamics and crowd and traffic control. In three dimensions, tracking several
targets becomes increasingly hard since optical occlusions are very likely,
i.e. two featureless targets frequently overlap for several frames. Occlusions
are particularly frequent in biological groups such as bird flocks, fish
schools, and insect swarms, a fact that has severely limited collective animal
behavior field studies in the past. This paper presents a 3D tracking method
that is robust in the case of severe occlusions. To ensure robustness, we adopt
a global optimization approach that works on all objects and frames at once. To
achieve practicality and scalability, we employ a divide and conquer
formulation, thanks to which the computational complexity of the problem is
reduced by orders of magnitude. We tested our algorithm with synthetic data,
with experimental data of bird flocks and insect swarms and with public
benchmark datasets, and show that our system yields high quality trajectories
for hundreds of moving targets with severe overlap. The results obtained on
very heterogeneous data show the potential applicability of our method to the
most diverse experimental situations.
| [
{
"version": "v1",
"created": "Tue, 7 May 2013 12:45:30 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Apr 2014 14:55:59 GMT"
},
{
"version": "v3",
"created": "Fri, 17 Apr 2015 16:36:59 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Attanasi",
"Alessandro",
""
],
[
"Cavagna",
"Andrea",
""
],
[
"Del Castello",
"Lorenzo",
""
],
[
"Giardina",
"Irene",
""
],
[
"Jelic",
"Asja",
""
],
[
"Melillo",
"Stefania",
""
],
[
"Parisi",
"Leonardo",
""
],
[
"Pellacini",
"Fabio",
""
],
[
"Shen",
"Edward",
""
],
[
"Silvestri",
"Edmondo",
""
],
[
"Viale",
"Massimiliano",
""
]
] | TITLE: GReTA - a novel Global and Recursive Tracking Algorithm in three
dimensions
ABSTRACT: Tracking multiple moving targets allows quantitative measure of the dynamic
behavior in systems as diverse as animal groups in biology, turbulence in fluid
dynamics and crowd and traffic control. In three dimensions, tracking several
targets becomes increasingly hard since optical occlusions are very likely,
i.e. two featureless targets frequently overlap for several frames. Occlusions
are particularly frequent in biological groups such as bird flocks, fish
schools, and insect swarms, a fact that has severely limited collective animal
behavior field studies in the past. This paper presents a 3D tracking method
that is robust in the case of severe occlusions. To ensure robustness, we adopt
a global optimization approach that works on all objects and frames at once. To
achieve practicality and scalability, we employ a divide and conquer
formulation, thanks to which the computational complexity of the problem is
reduced by orders of magnitude. We tested our algorithm with synthetic data,
with experimental data of bird flocks and insect swarms and with public
benchmark datasets, and show that our system yields high quality trajectories
for hundreds of moving targets with severe overlap. The results obtained on
very heterogeneous data show the potential applicability of our method to the
most diverse experimental situations.
| no_new_dataset | 0.950869 |
1408.5661 | Keisuke Yamazaki | Keisuke Yamazaki | Asymptotic Accuracy of Bayesian Estimation for a Single Latent Variable | 28 pages, 3 figures | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In data science and machine learning, hierarchical parametric models, such as
mixture models, are often used. They contain two kinds of variables: observable
variables, which represent the parts of the data that can be directly measured,
and latent variables, which represent the underlying processes that generate
the data. Although there has been an increase in research on the estimation
accuracy for observable variables, the theoretical analysis of estimating
latent variables has not been thoroughly investigated. In a previous study, we
determined the accuracy of a Bayes estimation for the joint probability of the
latent variables in a dataset, and we proved that the Bayes method is
asymptotically more accurate than the maximum-likelihood method. However, the
accuracy of the Bayes estimation for a single latent variable remains unknown.
In the present paper, we derive the asymptotic expansions of the error
functions, which are defined by the Kullback-Leibler divergence, for two types
of single-variable estimations when the statistical regularity is satisfied.
Our results indicate that the accuracies of the Bayes and maximum-likelihood
methods are asymptotically equivalent and clarify that the Bayes method is only
advantageous for multivariable estimations.
| [
{
"version": "v1",
"created": "Mon, 25 Aug 2014 04:44:53 GMT"
},
{
"version": "v2",
"created": "Tue, 27 Jan 2015 04:00:06 GMT"
},
{
"version": "v3",
"created": "Fri, 17 Apr 2015 06:59:26 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Yamazaki",
"Keisuke",
""
]
] | TITLE: Asymptotic Accuracy of Bayesian Estimation for a Single Latent Variable
ABSTRACT: In data science and machine learning, hierarchical parametric models, such as
mixture models, are often used. They contain two kinds of variables: observable
variables, which represent the parts of the data that can be directly measured,
and latent variables, which represent the underlying processes that generate
the data. Although there has been an increase in research on the estimation
accuracy for observable variables, the theoretical analysis of estimating
latent variables has not been thoroughly investigated. In a previous study, we
determined the accuracy of a Bayes estimation for the joint probability of the
latent variables in a dataset, and we proved that the Bayes method is
asymptotically more accurate than the maximum-likelihood method. However, the
accuracy of the Bayes estimation for a single latent variable remains unknown.
In the present paper, we derive the asymptotic expansions of the error
functions, which are defined by the Kullback-Leibler divergence, for two types
of single-variable estimations when the statistical regularity is satisfied.
Our results indicate that the accuracies of the Bayes and maximum-likelihood
methods are asymptotically equivalent and clarify that the Bayes method is only
advantageous for multivariable estimations.
| no_new_dataset | 0.947914 |
1412.5083 | Qiang Qiu | Qiang Qiu, Guillermo Sapiro, Alex Bronstein | Random Forests Can Hash | null | null | null | null | cs.CV cs.IR cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hash codes are a very efficient data representation needed to be able to cope
with the ever growing amounts of data. We introduce a random forest semantic
hashing scheme with information-theoretic code aggregation, showing for the
first time how random forest, a technique that together with deep learning have
shown spectacular results in classification, can also be extended to
large-scale retrieval. Traditional random forest fails to enforce the
consistency of hashes generated from each tree for the same class data, i.e.,
to preserve the underlying similarity, and it also lacks a principled way for
code aggregation across trees. We start with a simple hashing scheme, where
independently trained random trees in a forest are acting as hashing functions.
We the propose a subspace model as the splitting function, and show that it
enforces the hash consistency in a tree for data from the same class. We also
introduce an information-theoretic approach for aggregating codes of individual
trees into a single hash code, producing a near-optimal unique hash for each
class. Experiments on large-scale public datasets are presented, showing that
the proposed approach significantly outperforms state-of-the-art hashing
methods for retrieval tasks.
| [
{
"version": "v1",
"created": "Tue, 16 Dec 2014 17:02:18 GMT"
},
{
"version": "v2",
"created": "Tue, 24 Feb 2015 18:26:12 GMT"
},
{
"version": "v3",
"created": "Fri, 17 Apr 2015 01:00:24 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Qiu",
"Qiang",
""
],
[
"Sapiro",
"Guillermo",
""
],
[
"Bronstein",
"Alex",
""
]
] | TITLE: Random Forests Can Hash
ABSTRACT: Hash codes are a very efficient data representation needed to be able to cope
with the ever growing amounts of data. We introduce a random forest semantic
hashing scheme with information-theoretic code aggregation, showing for the
first time how random forest, a technique that together with deep learning have
shown spectacular results in classification, can also be extended to
large-scale retrieval. Traditional random forest fails to enforce the
consistency of hashes generated from each tree for the same class data, i.e.,
to preserve the underlying similarity, and it also lacks a principled way for
code aggregation across trees. We start with a simple hashing scheme, where
independently trained random trees in a forest are acting as hashing functions.
We the propose a subspace model as the splitting function, and show that it
enforces the hash consistency in a tree for data from the same class. We also
introduce an information-theoretic approach for aggregating codes of individual
trees into a single hash code, producing a near-optimal unique hash for each
class. Experiments on large-scale public datasets are presented, showing that
the proposed approach significantly outperforms state-of-the-art hashing
methods for retrieval tasks.
| no_new_dataset | 0.949435 |
1503.05787 | Tobias Isenberg | Maarten H. Everts, Henk Bekker, Jos B.T.M. Roerdink, and Tobias
Isenberg | Interactive Illustrative Line Styles and Line Style Transfer Functions
for Flow Visualization | Extended version of a short paper at Pacific Graphics 2011
(http://dx.doi.org/10.2312/PE/PG/PG2011short/105-110) | null | null | null | cs.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a flexible illustrative line style model for the visualization of
streamline data. Our model partitions view-oriented line strips into parallel
bands whose basic visual properties can be controlled independently. We thus
extend previous line stylization techniques specifically for visualization
purposes by allowing the parametrization of these bands based on the local line
data attributes. Moreover, our approach supports emphasis and abstraction by
introducing line style transfer functions that map local line attribute values
to complete line styles. With a flexible GPU implementation of this line style
model we enable the interactive exploration of visual representations of
streamlines. We demonstrate the effectiveness of our model by applying it to 3D
flow field datasets.
| [
{
"version": "v1",
"created": "Thu, 19 Mar 2015 14:53:55 GMT"
},
{
"version": "v2",
"created": "Fri, 17 Apr 2015 14:01:50 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Everts",
"Maarten H.",
""
],
[
"Bekker",
"Henk",
""
],
[
"Roerdink",
"Jos B. T. M.",
""
],
[
"Isenberg",
"Tobias",
""
]
] | TITLE: Interactive Illustrative Line Styles and Line Style Transfer Functions
for Flow Visualization
ABSTRACT: We present a flexible illustrative line style model for the visualization of
streamline data. Our model partitions view-oriented line strips into parallel
bands whose basic visual properties can be controlled independently. We thus
extend previous line stylization techniques specifically for visualization
purposes by allowing the parametrization of these bands based on the local line
data attributes. Moreover, our approach supports emphasis and abstraction by
introducing line style transfer functions that map local line attribute values
to complete line styles. With a flexible GPU implementation of this line style
model we enable the interactive exploration of visual representations of
streamlines. We demonstrate the effectiveness of our model by applying it to 3D
flow field datasets.
| no_new_dataset | 0.948394 |
1504.04531 | Nicolas Dobigeon | Laetitia Loncan, Luis B. Almeida, Jos\'e M. Bioucas-Dias, Xavier
Briottet, Jocelyn Chanussot, Nicolas Dobigeon, Sophie Fabre, Wenzhi Liao,
Giorgio A. Licciardi, Miguel Sim\~oes, Jean-Yves Tourneret, Miguel A.
Veganzones, Gemine Vivone, Qi Wei and Naoto Yokoya | Hyperspectral pansharpening: a review | null | null | null | null | cs.CV physics.data-an stat.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Pansharpening aims at fusing a panchromatic image with a multispectral one,
to generate an image with the high spatial resolution of the former and the
high spectral resolution of the latter. In the last decade, many algorithms
have been presented in the literature for pansharpening using multispectral
data. With the increasing availability of hyperspectral systems, these methods
are now being adapted to hyperspectral images. In this work, we compare new
pansharpening techniques designed for hyperspectral data with some of the state
of the art methods for multispectral pansharpening, which have been adapted for
hyperspectral data. Eleven methods from different classes (component
substitution, multiresolution analysis, hybrid, Bayesian and matrix
factorization) are analyzed. These methods are applied to three datasets and
their effectiveness and robustness are evaluated with widely used performance
indicators. In addition, all the pansharpening techniques considered in this
paper have been implemented in a MATLAB toolbox that is made available to the
community.
| [
{
"version": "v1",
"created": "Fri, 17 Apr 2015 15:07:11 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Loncan",
"Laetitia",
""
],
[
"Almeida",
"Luis B.",
""
],
[
"Bioucas-Dias",
"José M.",
""
],
[
"Briottet",
"Xavier",
""
],
[
"Chanussot",
"Jocelyn",
""
],
[
"Dobigeon",
"Nicolas",
""
],
[
"Fabre",
"Sophie",
""
],
[
"Liao",
"Wenzhi",
""
],
[
"Licciardi",
"Giorgio A.",
""
],
[
"Simões",
"Miguel",
""
],
[
"Tourneret",
"Jean-Yves",
""
],
[
"Veganzones",
"Miguel A.",
""
],
[
"Vivone",
"Gemine",
""
],
[
"Wei",
"Qi",
""
],
[
"Yokoya",
"Naoto",
""
]
] | TITLE: Hyperspectral pansharpening: a review
ABSTRACT: Pansharpening aims at fusing a panchromatic image with a multispectral one,
to generate an image with the high spatial resolution of the former and the
high spectral resolution of the latter. In the last decade, many algorithms
have been presented in the literature for pansharpening using multispectral
data. With the increasing availability of hyperspectral systems, these methods
are now being adapted to hyperspectral images. In this work, we compare new
pansharpening techniques designed for hyperspectral data with some of the state
of the art methods for multispectral pansharpening, which have been adapted for
hyperspectral data. Eleven methods from different classes (component
substitution, multiresolution analysis, hybrid, Bayesian and matrix
factorization) are analyzed. These methods are applied to three datasets and
their effectiveness and robustness are evaluated with widely used performance
indicators. In addition, all the pansharpening techniques considered in this
paper have been implemented in a MATLAB toolbox that is made available to the
community.
| no_new_dataset | 0.952175 |
1504.04548 | Simone Bianco | Simone Bianco, Claudio Cusano, Raimondo Schettini | Color Constancy Using CNNs | Accepted at DeepVision: Deep Learning in Computer Vision 2015 (CVPR
2015 workshop) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we describe a Convolutional Neural Network (CNN) to accurately
predict the scene illumination. Taking image patches as input, the CNN works in
the spatial domain without using hand-crafted features that are employed by
most previous methods. The network consists of one convolutional layer with max
pooling, one fully connected layer and three output nodes. Within the network
structure, feature learning and regression are integrated into one optimization
process, which leads to a more effective model for estimating scene
illumination. This approach achieves state-of-the-art performance on a standard
dataset of RAW images. Preliminary experiments on images with spatially varying
illumination demonstrate the stability of the local illuminant estimation
ability of our CNN.
| [
{
"version": "v1",
"created": "Fri, 17 Apr 2015 15:51:07 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Bianco",
"Simone",
""
],
[
"Cusano",
"Claudio",
""
],
[
"Schettini",
"Raimondo",
""
]
] | TITLE: Color Constancy Using CNNs
ABSTRACT: In this work we describe a Convolutional Neural Network (CNN) to accurately
predict the scene illumination. Taking image patches as input, the CNN works in
the spatial domain without using hand-crafted features that are employed by
most previous methods. The network consists of one convolutional layer with max
pooling, one fully connected layer and three output nodes. Within the network
structure, feature learning and regression are integrated into one optimization
process, which leads to a more effective model for estimating scene
illumination. This approach achieves state-of-the-art performance on a standard
dataset of RAW images. Preliminary experiments on images with spatially varying
illumination demonstrate the stability of the local illuminant estimation
ability of our CNN.
| no_new_dataset | 0.948489 |
1504.04588 | James-A. Goulet | James-A. Goulet | The Nataf-Beta Random Field Classifier: An Extension of the Beta
Conjugate Prior to Classification Problems | 17 pages, 4 figures, Submitted for publication in the Journal of
Machine Learning Research | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents the Nataf-Beta Random Field Classifier, a discriminative
approach that extends the applicability of the Beta conjugate prior to
classification problems. The approach's key feature is to model the probability
of a class conditional on attribute values as a random field whose marginals
are Beta distributed, and where the parameters of marginals are themselves
described by random fields. Although the classification accuracy of the
approach proposed does not statistically outperform the best accuracies
reported in the literature, it ranks among the top tier for the six benchmark
datasets tested. The Nataf-Beta Random Field Classifier is suited as a general
purpose classification approach for real-continuous and real-integer attribute
value problems.
| [
{
"version": "v1",
"created": "Fri, 17 Apr 2015 17:32:00 GMT"
}
] | 2015-04-20T00:00:00 | [
[
"Goulet",
"James-A.",
""
]
] | TITLE: The Nataf-Beta Random Field Classifier: An Extension of the Beta
Conjugate Prior to Classification Problems
ABSTRACT: This paper presents the Nataf-Beta Random Field Classifier, a discriminative
approach that extends the applicability of the Beta conjugate prior to
classification problems. The approach's key feature is to model the probability
of a class conditional on attribute values as a random field whose marginals
are Beta distributed, and where the parameters of marginals are themselves
described by random fields. Although the classification accuracy of the
approach proposed does not statistically outperform the best accuracies
reported in the literature, it ranks among the top tier for the six benchmark
datasets tested. The Nataf-Beta Random Field Classifier is suited as a general
purpose classification approach for real-continuous and real-integer attribute
value problems.
| no_new_dataset | 0.952397 |
1409.0940 | Haim Avron | Vikas Sindhwani and Haim Avron | High-performance Kernel Machines with Implicit Distributed Optimization
and Randomization | Work presented at MMDS 2014 (June 2014) and JSM 2014 | null | null | null | stat.ML cs.DC cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In order to fully utilize "big data", it is often required to use "big
models". Such models tend to grow with the complexity and size of the training
data, and do not make strong parametric assumptions upfront on the nature of
the underlying statistical dependencies. Kernel methods fit this need well, as
they constitute a versatile and principled statistical methodology for solving
a wide range of non-parametric modelling problems. However, their high
computational costs (in storage and time) pose a significant barrier to their
widespread adoption in big data applications.
We propose an algorithmic framework and high-performance implementation for
massive-scale training of kernel-based statistical models, based on combining
two key technical ingredients: (i) distributed general purpose convex
optimization, and (ii) the use of randomization to improve the scalability of
kernel methods. Our approach is based on a block-splitting variant of the
Alternating Directions Method of Multipliers, carefully reconfigured to handle
very large random feature matrices, while exploiting hybrid parallelism
typically found in modern clusters of multicore machines. Our implementation
supports a variety of statistical learning tasks by enabling several loss
functions, regularization schemes, kernels, and layers of randomized
approximations for both dense and sparse datasets, in a highly extensible
framework. We evaluate the ability of our framework to learn models on data
from applications, and provide a comparison against existing sequential and
parallel libraries.
| [
{
"version": "v1",
"created": "Wed, 3 Sep 2014 02:28:51 GMT"
},
{
"version": "v2",
"created": "Tue, 23 Dec 2014 21:38:15 GMT"
},
{
"version": "v3",
"created": "Thu, 16 Apr 2015 18:06:53 GMT"
}
] | 2015-04-17T00:00:00 | [
[
"Sindhwani",
"Vikas",
""
],
[
"Avron",
"Haim",
""
]
] | TITLE: High-performance Kernel Machines with Implicit Distributed Optimization
and Randomization
ABSTRACT: In order to fully utilize "big data", it is often required to use "big
models". Such models tend to grow with the complexity and size of the training
data, and do not make strong parametric assumptions upfront on the nature of
the underlying statistical dependencies. Kernel methods fit this need well, as
they constitute a versatile and principled statistical methodology for solving
a wide range of non-parametric modelling problems. However, their high
computational costs (in storage and time) pose a significant barrier to their
widespread adoption in big data applications.
We propose an algorithmic framework and high-performance implementation for
massive-scale training of kernel-based statistical models, based on combining
two key technical ingredients: (i) distributed general purpose convex
optimization, and (ii) the use of randomization to improve the scalability of
kernel methods. Our approach is based on a block-splitting variant of the
Alternating Directions Method of Multipliers, carefully reconfigured to handle
very large random feature matrices, while exploiting hybrid parallelism
typically found in modern clusters of multicore machines. Our implementation
supports a variety of statistical learning tasks by enabling several loss
functions, regularization schemes, kernels, and layers of randomized
approximations for both dense and sparse datasets, in a highly extensible
framework. We evaluate the ability of our framework to learn models on data
from applications, and provide a comparison against existing sequential and
parallel libraries.
| no_new_dataset | 0.940353 |
1502.05337 | Emiliano De Cristofaro | Julien Freudiger and Emiliano De Cristofaro and Alex Brito | Controlled Data Sharing for Collaborative Predictive Blacklisting | A preliminary version of this paper appears in DIMVA 2015. This is
the full version. arXiv admin note: substantial text overlap with
arXiv:1403.2123 | null | null | null | cs.CR cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Although sharing data across organizations is often advocated as a promising
way to enhance cybersecurity, collaborative initiatives are rarely put into
practice owing to confidentiality, trust, and liability challenges. In this
paper, we investigate whether collaborative threat mitigation can be realized
via a controlled data sharing approach, whereby organizations make informed
decisions as to whether or not, and how much, to share. Using appropriate
cryptographic tools, entities can estimate the benefits of collaboration and
agree on what to share in a privacy-preserving way, without having to disclose
their datasets. We focus on collaborative predictive blacklisting, i.e.,
forecasting attack sources based on one's logs and those contributed by other
organizations. We study the impact of different sharing strategies by
experimenting on a real-world dataset of two billion suspicious IP addresses
collected from Dshield over two months. We find that controlled data sharing
yields up to 105% accuracy improvement on average, while also reducing the
false positive rate.
| [
{
"version": "v1",
"created": "Wed, 18 Feb 2015 18:48:56 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Apr 2015 09:09:16 GMT"
}
] | 2015-04-17T00:00:00 | [
[
"Freudiger",
"Julien",
""
],
[
"De Cristofaro",
"Emiliano",
""
],
[
"Brito",
"Alex",
""
]
] | TITLE: Controlled Data Sharing for Collaborative Predictive Blacklisting
ABSTRACT: Although sharing data across organizations is often advocated as a promising
way to enhance cybersecurity, collaborative initiatives are rarely put into
practice owing to confidentiality, trust, and liability challenges. In this
paper, we investigate whether collaborative threat mitigation can be realized
via a controlled data sharing approach, whereby organizations make informed
decisions as to whether or not, and how much, to share. Using appropriate
cryptographic tools, entities can estimate the benefits of collaboration and
agree on what to share in a privacy-preserving way, without having to disclose
their datasets. We focus on collaborative predictive blacklisting, i.e.,
forecasting attack sources based on one's logs and those contributed by other
organizations. We study the impact of different sharing strategies by
experimenting on a real-world dataset of two billion suspicious IP addresses
collected from Dshield over two months. We find that controlled data sharing
yields up to 105% accuracy improvement on average, while also reducing the
false positive rate.
| no_new_dataset | 0.847968 |
1503.01224 | Chunhua Shen | Peng Wang, Yuanzhouhan Cao, Chunhua Shen, Lingqiao Liu, Heng Tao Shen | Temporal Pyramid Pooling Based Convolutional Neural Networks for Action
Recognition | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Encouraged by the success of Convolutional Neural Networks (CNNs) in image
classification, recently much effort is spent on applying CNNs to video based
action recognition problems. One challenge is that video contains a varying
number of frames which is incompatible to the standard input format of CNNs.
Existing methods handle this issue either by directly sampling a fixed number
of frames or bypassing this issue by introducing a 3D convolutional layer which
conducts convolution in spatial-temporal domain.
To solve this issue, here we propose a novel network structure which allows
an arbitrary number of frames as the network input. The key of our solution is
to introduce a module consisting of an encoding layer and a temporal pyramid
pooling layer. The encoding layer maps the activation from previous layers to a
feature vector suitable for pooling while the temporal pyramid pooling layer
converts multiple frame-level activations into a fixed-length video-level
representation. In addition, we adopt a feature concatenation layer which
combines appearance information and motion information. Compared with the frame
sampling strategy, our method avoids the risk of missing any important frames.
Compared with the 3D convolutional method which requires a huge video dataset
for network training, our model can be learned on a small target dataset
because we can leverage the off-the-shelf image-level CNN for model parameter
initialization. Experiments on two challenging datasets, Hollywood2 and HMDB51,
demonstrate that our method achieves superior performance over state-of-the-art
methods while requiring much fewer training data.
| [
{
"version": "v1",
"created": "Wed, 4 Mar 2015 05:18:18 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Apr 2015 12:18:46 GMT"
}
] | 2015-04-17T00:00:00 | [
[
"Wang",
"Peng",
""
],
[
"Cao",
"Yuanzhouhan",
""
],
[
"Shen",
"Chunhua",
""
],
[
"Liu",
"Lingqiao",
""
],
[
"Shen",
"Heng Tao",
""
]
] | TITLE: Temporal Pyramid Pooling Based Convolutional Neural Networks for Action
Recognition
ABSTRACT: Encouraged by the success of Convolutional Neural Networks (CNNs) in image
classification, recently much effort is spent on applying CNNs to video based
action recognition problems. One challenge is that video contains a varying
number of frames which is incompatible to the standard input format of CNNs.
Existing methods handle this issue either by directly sampling a fixed number
of frames or bypassing this issue by introducing a 3D convolutional layer which
conducts convolution in spatial-temporal domain.
To solve this issue, here we propose a novel network structure which allows
an arbitrary number of frames as the network input. The key of our solution is
to introduce a module consisting of an encoding layer and a temporal pyramid
pooling layer. The encoding layer maps the activation from previous layers to a
feature vector suitable for pooling while the temporal pyramid pooling layer
converts multiple frame-level activations into a fixed-length video-level
representation. In addition, we adopt a feature concatenation layer which
combines appearance information and motion information. Compared with the frame
sampling strategy, our method avoids the risk of missing any important frames.
Compared with the 3D convolutional method which requires a huge video dataset
for network training, our model can be learned on a small target dataset
because we can leverage the off-the-shelf image-level CNN for model parameter
initialization. Experiments on two challenging datasets, Hollywood2 and HMDB51,
demonstrate that our method achieves superior performance over state-of-the-art
methods while requiring much fewer training data.
| no_new_dataset | 0.951639 |
1504.03641 | Sergey Zagoruyko | Sergey Zagoruyko and Nikos Komodakis | Learning to Compare Image Patches via Convolutional Neural Networks | CVPR 2015 | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we show how to learn directly from image data (i.e., without
resorting to manually-designed features) a general similarity function for
comparing image patches, which is a task of fundamental importance for many
computer vision problems. To encode such a function, we opt for a CNN-based
model that is trained to account for a wide variety of changes in image
appearance. To that end, we explore and study multiple neural network
architectures, which are specifically adapted to this task. We show that such
an approach can significantly outperform the state-of-the-art on several
problems and benchmark datasets.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 17:53:51 GMT"
}
] | 2015-04-17T00:00:00 | [
[
"Zagoruyko",
"Sergey",
""
],
[
"Komodakis",
"Nikos",
""
]
] | TITLE: Learning to Compare Image Patches via Convolutional Neural Networks
ABSTRACT: In this paper we show how to learn directly from image data (i.e., without
resorting to manually-designed features) a general similarity function for
comparing image patches, which is a task of fundamental importance for many
computer vision problems. To encode such a function, we opt for a CNN-based
model that is trained to account for a wide variety of changes in image
appearance. To that end, we explore and study multiple neural network
architectures, which are specifically adapted to this task. We show that such
an approach can significantly outperform the state-of-the-art on several
problems and benchmark datasets.
| no_new_dataset | 0.949576 |
1504.04054 | Xin Yuan | Yunchen Pu, Xin Yuan, Lawrence Carin | A Generative Model for Deep Convolutional Learning | 3 pages, 1 figure, ICLR workshop | null | null | null | stat.ML cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A generative model is developed for deep (multi-layered) convolutional
dictionary learning. A novel probabilistic pooling operation is integrated into
the deep model, yielding efficient bottom-up (pretraining) and top-down
(refinement) probabilistic learning. Experimental results demonstrate powerful
capabilities of the model to learn multi-layer features from images, and
excellent classification results are obtained on the MNIST and Caltech 101
datasets.
| [
{
"version": "v1",
"created": "Wed, 15 Apr 2015 21:31:58 GMT"
}
] | 2015-04-17T00:00:00 | [
[
"Pu",
"Yunchen",
""
],
[
"Yuan",
"Xin",
""
],
[
"Carin",
"Lawrence",
""
]
] | TITLE: A Generative Model for Deep Convolutional Learning
ABSTRACT: A generative model is developed for deep (multi-layered) convolutional
dictionary learning. A novel probabilistic pooling operation is integrated into
the deep model, yielding efficient bottom-up (pretraining) and top-down
(refinement) probabilistic learning. Experimental results demonstrate powerful
capabilities of the model to learn multi-layer features from images, and
excellent classification results are obtained on the MNIST and Caltech 101
datasets.
| no_new_dataset | 0.946892 |
1504.04208 | Andrea Scharnhorst | Rob Koopman, Shenghui Wang, Andrea Scharnhorst | Contextualization of topics - browsing through terms, authors, journals
and cluster allocations | proceedings of the ISSI 2015 conference (accepted) | null | null | null | cs.DL cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper builds on an innovative Information Retrieval tool, Ariadne. The
tool has been developed as an interactive network visualization and browsing
tool for large-scale bibliographic databases. It basically allows to gain
insights into a topic by contextualizing a search query (Koopman et al., 2015).
In this paper, we apply the Ariadne tool to a far smaller dataset of 111,616
documents in astronomy and astrophysics. Labeled as the Berlin dataset, this
data have been used by several research teams to apply and later compare
different clustering algorithms. The quest for this team effort is how to
delineate topics. This paper contributes to this challenge in two different
ways. First, we produce one of the different cluster solution and second, we
use Ariadne (the method behind it, and the interface - called LittleAriadne) to
display cluster solutions of the different group members. By providing a tool
that allows the visual inspection of the similarity of article clusters
produced by different algorithms, we present a complementary approach to other
possible means of comparison. More particular, we discuss how we can - with
LittleAriadne - browse through the network of topical terms, authors, journals
and cluster solutions in the Berlin dataset and compare cluster solutions as
well as see their context.
| [
{
"version": "v1",
"created": "Thu, 16 Apr 2015 12:38:10 GMT"
}
] | 2015-04-17T00:00:00 | [
[
"Koopman",
"Rob",
""
],
[
"Wang",
"Shenghui",
""
],
[
"Scharnhorst",
"Andrea",
""
]
] | TITLE: Contextualization of topics - browsing through terms, authors, journals
and cluster allocations
ABSTRACT: This paper builds on an innovative Information Retrieval tool, Ariadne. The
tool has been developed as an interactive network visualization and browsing
tool for large-scale bibliographic databases. It basically allows to gain
insights into a topic by contextualizing a search query (Koopman et al., 2015).
In this paper, we apply the Ariadne tool to a far smaller dataset of 111,616
documents in astronomy and astrophysics. Labeled as the Berlin dataset, this
data have been used by several research teams to apply and later compare
different clustering algorithms. The quest for this team effort is how to
delineate topics. This paper contributes to this challenge in two different
ways. First, we produce one of the different cluster solution and second, we
use Ariadne (the method behind it, and the interface - called LittleAriadne) to
display cluster solutions of the different group members. By providing a tool
that allows the visual inspection of the similarity of article clusters
produced by different algorithms, we present a complementary approach to other
possible means of comparison. More particular, we discuss how we can - with
LittleAriadne - browse through the network of topical terms, authors, journals
and cluster solutions in the Berlin dataset and compare cluster solutions as
well as see their context.
| no_new_dataset | 0.888227 |
1406.5670 | Zhirong Wu | Zhirong Wu, Shuran Song, Aditya Khosla, Fisher Yu, Linguang Zhang,
Xiaoou Tang, Jianxiong Xiao | 3D ShapeNets: A Deep Representation for Volumetric Shapes | to be appeared in CVPR 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | 3D shape is a crucial but heavily underutilized cue in today's computer
vision systems, mostly due to the lack of a good generic shape representation.
With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft
Kinect), it is becoming increasingly important to have a powerful 3D shape
representation in the loop. Apart from category recognition, recovering full 3D
shapes from view-based 2.5D depth maps is also a critical part of visual
understanding. To this end, we propose to represent a geometric 3D shape as a
probability distribution of binary variables on a 3D voxel grid, using a
Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the
distribution of complex 3D shapes across different object categories and
arbitrary poses from raw CAD data, and discovers hierarchical compositional
part representations automatically. It naturally supports joint object
recognition and shape completion from 2.5D depth maps, and it enables active
object recognition through view planning. To train our 3D deep learning model,
we construct ModelNet -- a large-scale 3D CAD model dataset. Extensive
experiments show that our 3D deep representation enables significant
performance improvement over the-state-of-the-arts in a variety of tasks.
| [
{
"version": "v1",
"created": "Sun, 22 Jun 2014 03:31:52 GMT"
},
{
"version": "v2",
"created": "Mon, 1 Sep 2014 04:59:49 GMT"
},
{
"version": "v3",
"created": "Wed, 15 Apr 2015 16:46:05 GMT"
}
] | 2015-04-16T00:00:00 | [
[
"Wu",
"Zhirong",
""
],
[
"Song",
"Shuran",
""
],
[
"Khosla",
"Aditya",
""
],
[
"Yu",
"Fisher",
""
],
[
"Zhang",
"Linguang",
""
],
[
"Tang",
"Xiaoou",
""
],
[
"Xiao",
"Jianxiong",
""
]
] | TITLE: 3D ShapeNets: A Deep Representation for Volumetric Shapes
ABSTRACT: 3D shape is a crucial but heavily underutilized cue in today's computer
vision systems, mostly due to the lack of a good generic shape representation.
With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft
Kinect), it is becoming increasingly important to have a powerful 3D shape
representation in the loop. Apart from category recognition, recovering full 3D
shapes from view-based 2.5D depth maps is also a critical part of visual
understanding. To this end, we propose to represent a geometric 3D shape as a
probability distribution of binary variables on a 3D voxel grid, using a
Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the
distribution of complex 3D shapes across different object categories and
arbitrary poses from raw CAD data, and discovers hierarchical compositional
part representations automatically. It naturally supports joint object
recognition and shape completion from 2.5D depth maps, and it enables active
object recognition through view planning. To train our 3D deep learning model,
we construct ModelNet -- a large-scale 3D CAD model dataset. Extensive
experiments show that our 3D deep representation enables significant
performance improvement over the-state-of-the-arts in a variety of tasks.
| new_dataset | 0.902352 |
1410.6910 | Thomas Kreuz | Thomas Kreuz and Mario Mulansky and Nebojsa Bozanic | SPIKY: A graphical user interface for monitoring spike train synchrony | 15 pages, 7 figures | null | null | null | physics.data-an cs.MS cs.SE physics.bio-ph physics.med-ph q-bio.NC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Techniques for recording large-scale neuronal spiking activity are developing
very fast. This leads to an increasing demand for algorithms capable of
analyzing large amounts of experimental spike train data. One of the most
crucial and demanding tasks is the identification of similarity patterns with a
very high temporal resolution and across different spatial scales. To address
this task, in recent years three time-resolved measures of spike train
synchrony have been proposed, the ISI-distance, the SPIKE-distance, and event
synchronization. The Matlab source codes for calculating and visualizing these
measures have been made publicly available. However, due to the many different
possible representations of the results the use of these codes is rather
complicated and their application requires some basic knowledge of Matlab. Thus
it became desirable to provide a more user-friendly and interactive interface.
Here we address this need and present SPIKY, a graphical user interface which
facilitates the application of time-resolved measures of spike train synchrony
to both simulated and real data. SPIKY includes implementations of the
ISI-distance, the SPIKE-distance and SPIKE-synchronization (an improved and
simplified extension of event synchronization) which have been optimized with
respect to computation speed and memory demand. It also comprises a spike train
generator and an event detector which makes it capable of analyzing continuous
data. Finally, the SPIKY package includes additional complementary programs
aimed at the analysis of large numbers of datasets and the estimation of
significance levels.
| [
{
"version": "v1",
"created": "Sat, 25 Oct 2014 11:02:26 GMT"
},
{
"version": "v2",
"created": "Sat, 24 Jan 2015 22:29:28 GMT"
},
{
"version": "v3",
"created": "Wed, 15 Apr 2015 16:40:11 GMT"
}
] | 2015-04-16T00:00:00 | [
[
"Kreuz",
"Thomas",
""
],
[
"Mulansky",
"Mario",
""
],
[
"Bozanic",
"Nebojsa",
""
]
] | TITLE: SPIKY: A graphical user interface for monitoring spike train synchrony
ABSTRACT: Techniques for recording large-scale neuronal spiking activity are developing
very fast. This leads to an increasing demand for algorithms capable of
analyzing large amounts of experimental spike train data. One of the most
crucial and demanding tasks is the identification of similarity patterns with a
very high temporal resolution and across different spatial scales. To address
this task, in recent years three time-resolved measures of spike train
synchrony have been proposed, the ISI-distance, the SPIKE-distance, and event
synchronization. The Matlab source codes for calculating and visualizing these
measures have been made publicly available. However, due to the many different
possible representations of the results the use of these codes is rather
complicated and their application requires some basic knowledge of Matlab. Thus
it became desirable to provide a more user-friendly and interactive interface.
Here we address this need and present SPIKY, a graphical user interface which
facilitates the application of time-resolved measures of spike train synchrony
to both simulated and real data. SPIKY includes implementations of the
ISI-distance, the SPIKE-distance and SPIKE-synchronization (an improved and
simplified extension of event synchronization) which have been optimized with
respect to computation speed and memory demand. It also comprises a spike train
generator and an event detector which makes it capable of analyzing continuous
data. Finally, the SPIKY package includes additional complementary programs
aimed at the analysis of large numbers of datasets and the estimation of
significance levels.
| no_new_dataset | 0.942242 |
1412.6596 | Scott Reed | Scott Reed, Honglak Lee, Dragomir Anguelov, Christian Szegedy, Dumitru
Erhan, Andrew Rabinovich | Training Deep Neural Networks on Noisy Labels with Bootstrapping | null | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current state-of-the-art deep learning systems for visual object recognition
and detection use purely supervised training with regularization such as
dropout to avoid overfitting. The performance depends critically on the amount
of labeled examples, and in current practice the labels are assumed to be
unambiguous and accurate. However, this assumption often does not hold; e.g. in
recognition, class labels may be missing; in detection, objects in the image
may not be localized; and in general, the labeling may be subjective. In this
work we propose a generic way to handle noisy and incomplete labeling by
augmenting the prediction objective with a notion of consistency. We consider a
prediction consistent if the same prediction is made given similar percepts,
where the notion of similarity is between deep network features computed from
the input data. In experiments we demonstrate that our approach yields
substantial robustness to label noise on several datasets. On MNIST handwritten
digits, we show that our model is robust to label corruption. On the Toronto
Face Database, we show that our model handles well the case of subjective
labels in emotion recognition, achieving state-of-the- art results, and can
also benefit from unlabeled face images with no modification to our method. On
the ILSVRC2014 detection challenge data, we show that our approach extends to
very deep networks, high resolution images and structured outputs, and results
in improved scalable detection.
| [
{
"version": "v1",
"created": "Sat, 20 Dec 2014 04:11:33 GMT"
},
{
"version": "v2",
"created": "Sat, 7 Feb 2015 22:30:39 GMT"
},
{
"version": "v3",
"created": "Wed, 15 Apr 2015 19:48:37 GMT"
}
] | 2015-04-16T00:00:00 | [
[
"Reed",
"Scott",
""
],
[
"Lee",
"Honglak",
""
],
[
"Anguelov",
"Dragomir",
""
],
[
"Szegedy",
"Christian",
""
],
[
"Erhan",
"Dumitru",
""
],
[
"Rabinovich",
"Andrew",
""
]
] | TITLE: Training Deep Neural Networks on Noisy Labels with Bootstrapping
ABSTRACT: Current state-of-the-art deep learning systems for visual object recognition
and detection use purely supervised training with regularization such as
dropout to avoid overfitting. The performance depends critically on the amount
of labeled examples, and in current practice the labels are assumed to be
unambiguous and accurate. However, this assumption often does not hold; e.g. in
recognition, class labels may be missing; in detection, objects in the image
may not be localized; and in general, the labeling may be subjective. In this
work we propose a generic way to handle noisy and incomplete labeling by
augmenting the prediction objective with a notion of consistency. We consider a
prediction consistent if the same prediction is made given similar percepts,
where the notion of similarity is between deep network features computed from
the input data. In experiments we demonstrate that our approach yields
substantial robustness to label noise on several datasets. On MNIST handwritten
digits, we show that our model is robust to label corruption. On the Toronto
Face Database, we show that our model handles well the case of subjective
labels in emotion recognition, achieving state-of-the- art results, and can
also benefit from unlabeled face images with no modification to our method. On
the ILSVRC2014 detection challenge data, we show that our approach extends to
very deep networks, high resolution images and structured outputs, and results
in improved scalable detection.
| no_new_dataset | 0.944382 |
1404.3708 | Yuxiao Dong | Yuxiao Dong, Jie Tang, Nitesh Chawla, Tiancheng Lou, Yang Yang, Bai
Wang | Inferring Social Status and Rich Club Effects in Enterprise
Communication Networks | 13 pages, 4 figures | PLoS ONE 10(3): e0119446. 2015 | 10.1371/journal.pone.0119446 | null | cs.SI cs.AI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Social status, defined as the relative rank or position that an individual
holds in a social hierarchy, is known to be among the most important motivating
forces in social behaviors. In this paper, we consider the notion of status
from the perspective of a position or title held by a person in an enterprise.
We study the intersection of social status and social networks in an
enterprise. We study whether enterprise communication logs can help reveal how
social interactions and individual status manifest themselves in social
networks. To that end, we use two enterprise datasets with three communication
channels --- voice call, short message, and email --- to demonstrate the
social-behavioral differences among individuals with different status. We have
several interesting findings and based on these findings we also develop a
model to predict social status. On the individual level, high-status
individuals are more likely to be spanned as structural holes by linking to
people in parts of the enterprise networks that are otherwise not well
connected to one another. On the community level, the principle of homophily,
social balance and clique theory generally indicate a "rich club" maintained by
high-status individuals, in the sense that this community is much more
connected, balanced and dense. Our model can predict social status of
individuals with 93% accuracy.
| [
{
"version": "v1",
"created": "Mon, 14 Apr 2014 19:32:08 GMT"
},
{
"version": "v2",
"created": "Tue, 15 Apr 2014 01:25:26 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Apr 2014 20:12:21 GMT"
},
{
"version": "v4",
"created": "Tue, 14 Apr 2015 19:44:47 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Dong",
"Yuxiao",
""
],
[
"Tang",
"Jie",
""
],
[
"Chawla",
"Nitesh",
""
],
[
"Lou",
"Tiancheng",
""
],
[
"Yang",
"Yang",
""
],
[
"Wang",
"Bai",
""
]
] | TITLE: Inferring Social Status and Rich Club Effects in Enterprise
Communication Networks
ABSTRACT: Social status, defined as the relative rank or position that an individual
holds in a social hierarchy, is known to be among the most important motivating
forces in social behaviors. In this paper, we consider the notion of status
from the perspective of a position or title held by a person in an enterprise.
We study the intersection of social status and social networks in an
enterprise. We study whether enterprise communication logs can help reveal how
social interactions and individual status manifest themselves in social
networks. To that end, we use two enterprise datasets with three communication
channels --- voice call, short message, and email --- to demonstrate the
social-behavioral differences among individuals with different status. We have
several interesting findings and based on these findings we also develop a
model to predict social status. On the individual level, high-status
individuals are more likely to be spanned as structural holes by linking to
people in parts of the enterprise networks that are otherwise not well
connected to one another. On the community level, the principle of homophily,
social balance and clique theory generally indicate a "rich club" maintained by
high-status individuals, in the sense that this community is much more
connected, balanced and dense. Our model can predict social status of
individuals with 93% accuracy.
| no_new_dataset | 0.945701 |
1412.0623 | Sean Bell | Sean Bell and Paul Upchurch and Noah Snavely and Kavita Bala | Material Recognition in the Wild with the Materials in Context Database | CVPR 2015. Sean Bell and Paul Upchurch contributed equally | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recognizing materials in real-world images is a challenging task. Real-world
materials have rich surface texture, geometry, lighting conditions, and
clutter, which combine to make the problem particularly difficult. In this
paper, we introduce a new, large-scale, open dataset of materials in the wild,
the Materials in Context Database (MINC), and combine this dataset with deep
learning to achieve material recognition and segmentation of images in the
wild.
MINC is an order of magnitude larger than previous material databases, while
being more diverse and well-sampled across its 23 categories. Using MINC, we
train convolutional neural networks (CNNs) for two tasks: classifying materials
from patches, and simultaneous material recognition and segmentation in full
images. For patch-based classification on MINC we found that the best
performing CNN architectures can achieve 85.2% mean class accuracy. We convert
these trained CNN classifiers into an efficient fully convolutional framework
combined with a fully connected conditional random field (CRF) to predict the
material at every pixel in an image, achieving 73.1% mean class accuracy. Our
experiments demonstrate that having a large, well-sampled dataset such as MINC
is crucial for real-world material recognition and segmentation.
| [
{
"version": "v1",
"created": "Mon, 1 Dec 2014 20:11:44 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Apr 2015 05:29:32 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Bell",
"Sean",
""
],
[
"Upchurch",
"Paul",
""
],
[
"Snavely",
"Noah",
""
],
[
"Bala",
"Kavita",
""
]
] | TITLE: Material Recognition in the Wild with the Materials in Context Database
ABSTRACT: Recognizing materials in real-world images is a challenging task. Real-world
materials have rich surface texture, geometry, lighting conditions, and
clutter, which combine to make the problem particularly difficult. In this
paper, we introduce a new, large-scale, open dataset of materials in the wild,
the Materials in Context Database (MINC), and combine this dataset with deep
learning to achieve material recognition and segmentation of images in the
wild.
MINC is an order of magnitude larger than previous material databases, while
being more diverse and well-sampled across its 23 categories. Using MINC, we
train convolutional neural networks (CNNs) for two tasks: classifying materials
from patches, and simultaneous material recognition and segmentation in full
images. For patch-based classification on MINC we found that the best
performing CNN architectures can achieve 85.2% mean class accuracy. We convert
these trained CNN classifiers into an efficient fully convolutional framework
combined with a fully connected conditional random field (CRF) to predict the
material at every pixel in an image, achieving 73.1% mean class accuracy. Our
experiments demonstrate that having a large, well-sampled dataset such as MINC
is crucial for real-world material recognition and segmentation.
| new_dataset | 0.962883 |
1412.2306 | Andrej Karpathy | Andrej Karpathy, Li Fei-Fei | Deep Visual-Semantic Alignments for Generating Image Descriptions | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a model that generates natural language descriptions of images and
their regions. Our approach leverages datasets of images and their sentence
descriptions to learn about the inter-modal correspondences between language
and visual data. Our alignment model is based on a novel combination of
Convolutional Neural Networks over image regions, bidirectional Recurrent
Neural Networks over sentences, and a structured objective that aligns the two
modalities through a multimodal embedding. We then describe a Multimodal
Recurrent Neural Network architecture that uses the inferred alignments to
learn to generate novel descriptions of image regions. We demonstrate that our
alignment model produces state of the art results in retrieval experiments on
Flickr8K, Flickr30K and MSCOCO datasets. We then show that the generated
descriptions significantly outperform retrieval baselines on both full images
and on a new dataset of region-level annotations.
| [
{
"version": "v1",
"created": "Sun, 7 Dec 2014 02:36:07 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Apr 2015 05:02:53 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Karpathy",
"Andrej",
""
],
[
"Fei-Fei",
"Li",
""
]
] | TITLE: Deep Visual-Semantic Alignments for Generating Image Descriptions
ABSTRACT: We present a model that generates natural language descriptions of images and
their regions. Our approach leverages datasets of images and their sentence
descriptions to learn about the inter-modal correspondences between language
and visual data. Our alignment model is based on a novel combination of
Convolutional Neural Networks over image regions, bidirectional Recurrent
Neural Networks over sentences, and a structured objective that aligns the two
modalities through a multimodal embedding. We then describe a Multimodal
Recurrent Neural Network architecture that uses the inferred alignments to
learn to generate novel descriptions of image regions. We demonstrate that our
alignment model produces state of the art results in retrieval experiments on
Flickr8K, Flickr30K and MSCOCO datasets. We then show that the generated
descriptions significantly outperform retrieval baselines on both full images
and on a new dataset of region-level annotations.
| new_dataset | 0.951323 |
1412.3709 | Abel Gonzalez-Garcia | Abel Gonzalez-Garcia, Alexander Vezhnevets, Vittorio Ferrari | An active search strategy for efficient object class detection | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Object class detectors typically apply a window classifier to all the windows
in a large set, either in a sliding window manner or using object proposals. In
this paper, we develop an active search strategy that sequentially chooses the
next window to evaluate based on all the information gathered before. This
results in a substantial reduction in the number of classifier evaluations and
in a more elegant approach in general. Our search strategy is guided by two
forces. First, we exploit context as the statistical relation between the
appearance of a window and its location relative to the object, as observed in
the training set. This enables to jump across distant regions in the image
(e.g. observing a sky region suggests that cars might be far below) and is done
efficiently in a Random Forest framework. Second, we exploit the score of the
classifier to attract the search to promising areas surrounding a highly scored
window, and to keep away from areas near low scored ones. Our search strategy
can be applied on top of any classifier as it treats it as a black-box. In
experiments with R-CNN on the challenging SUN2012 dataset, our method matches
the detection accuracy of evaluating all windows independently, while
evaluating 9x fewer windows.
| [
{
"version": "v1",
"created": "Thu, 11 Dec 2014 16:23:38 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Apr 2015 11:29:51 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Gonzalez-Garcia",
"Abel",
""
],
[
"Vezhnevets",
"Alexander",
""
],
[
"Ferrari",
"Vittorio",
""
]
] | TITLE: An active search strategy for efficient object class detection
ABSTRACT: Object class detectors typically apply a window classifier to all the windows
in a large set, either in a sliding window manner or using object proposals. In
this paper, we develop an active search strategy that sequentially chooses the
next window to evaluate based on all the information gathered before. This
results in a substantial reduction in the number of classifier evaluations and
in a more elegant approach in general. Our search strategy is guided by two
forces. First, we exploit context as the statistical relation between the
appearance of a window and its location relative to the object, as observed in
the training set. This enables to jump across distant regions in the image
(e.g. observing a sky region suggests that cars might be far below) and is done
efficiently in a Random Forest framework. Second, we exploit the score of the
classifier to attract the search to promising areas surrounding a highly scored
window, and to keep away from areas near low scored ones. Our search strategy
can be applied on top of any classifier as it treats it as a black-box. In
experiments with R-CNN on the challenging SUN2012 dataset, our method matches
the detection accuracy of evaluating all windows independently, while
evaluating 9x fewer windows.
| no_new_dataset | 0.951684 |
1501.00100 | Marco Gramaglia | Marco Gramaglia and Marco Fiore | On the anonymizability of mobile traffic datasets | null | null | null | null | cs.CY cs.CR | http://creativecommons.org/licenses/by-nc-sa/3.0/ | Preserving user privacy is paramount when it comes to publicly disclosed
datasets that contain fine-grained data about large populations. The problem is
especially critical in the case of mobile traffic datasets collected by
cellular operators, as they feature elevate subscriber trajectory uniqueness
and they are resistant to anonymization through spatiotemporal generalization.
In this work, we investigate the $k$-anonymizability of trajectories in two
large-scale mobile traffic datasets, by means of a novel dedicated measure. Our
results are in agreement with those of previous analyses, however they also
provide additional insights on the reasons behind the poor anonimizability of
mobile traffic datasets. As such, our study is a step forward in the direction
of a more robust dataset anonymization.
| [
{
"version": "v1",
"created": "Wed, 31 Dec 2014 09:53:31 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Apr 2015 09:35:41 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Gramaglia",
"Marco",
""
],
[
"Fiore",
"Marco",
""
]
] | TITLE: On the anonymizability of mobile traffic datasets
ABSTRACT: Preserving user privacy is paramount when it comes to publicly disclosed
datasets that contain fine-grained data about large populations. The problem is
especially critical in the case of mobile traffic datasets collected by
cellular operators, as they feature elevate subscriber trajectory uniqueness
and they are resistant to anonymization through spatiotemporal generalization.
In this work, we investigate the $k$-anonymizability of trajectories in two
large-scale mobile traffic datasets, by means of a novel dedicated measure. Our
results are in agreement with those of previous analyses, however they also
provide additional insights on the reasons behind the poor anonimizability of
mobile traffic datasets. As such, our study is a step forward in the direction
of a more robust dataset anonymization.
| no_new_dataset | 0.952264 |
1502.01916 | Daniel Lemire | Wayne Xin Zhao, Xudong Zhang, Daniel Lemire, Dongdong Shan, Jian-Yun
Nie, Hongfei Yan, Ji-Rong Wen | A General SIMD-based Approach to Accelerating Compression Algorithms | null | ACM Trans. Inf. Syst. 33, 3, Article 15 (March 2015) | 10.1145/2735629 | null | cs.IR | http://creativecommons.org/licenses/by/3.0/ | Compression algorithms are important for data oriented tasks, especially in
the era of Big Data. Modern processors equipped with powerful SIMD instruction
sets, provide us an opportunity for achieving better compression performance.
Previous research has shown that SIMD-based optimizations can multiply decoding
speeds. Following these pioneering studies, we propose a general approach to
accelerate compression algorithms. By instantiating the approach, we have
developed several novel integer compression algorithms, called Group-Simple,
Group-Scheme, Group-AFOR, and Group-PFD, and implemented their corresponding
vectorized versions. We evaluate the proposed algorithms on two public TREC
datasets, a Wikipedia dataset and a Twitter dataset. With competitive
compression ratios and encoding speeds, our SIMD-based algorithms outperform
state-of-the-art non-vectorized algorithms with respect to decoding speeds.
| [
{
"version": "v1",
"created": "Fri, 6 Feb 2015 15:11:47 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Zhao",
"Wayne Xin",
""
],
[
"Zhang",
"Xudong",
""
],
[
"Lemire",
"Daniel",
""
],
[
"Shan",
"Dongdong",
""
],
[
"Nie",
"Jian-Yun",
""
],
[
"Yan",
"Hongfei",
""
],
[
"Wen",
"Ji-Rong",
""
]
] | TITLE: A General SIMD-based Approach to Accelerating Compression Algorithms
ABSTRACT: Compression algorithms are important for data oriented tasks, especially in
the era of Big Data. Modern processors equipped with powerful SIMD instruction
sets, provide us an opportunity for achieving better compression performance.
Previous research has shown that SIMD-based optimizations can multiply decoding
speeds. Following these pioneering studies, we propose a general approach to
accelerate compression algorithms. By instantiating the approach, we have
developed several novel integer compression algorithms, called Group-Simple,
Group-Scheme, Group-AFOR, and Group-PFD, and implemented their corresponding
vectorized versions. We evaluate the proposed algorithms on two public TREC
datasets, a Wikipedia dataset and a Twitter dataset. With competitive
compression ratios and encoding speeds, our SIMD-based algorithms outperform
state-of-the-art non-vectorized algorithms with respect to decoding speeds.
| no_new_dataset | 0.934275 |
1504.03154 | Carlo Ciliberto | Giulia Pasquale, Carlo Ciliberto, Francesca Odone, Lorenzo Rosasco and
Lorenzo Natale | Real-world Object Recognition with Off-the-shelf Deep Conv Nets: How
Many Objects can iCub Learn? | 18 pages, 9 figures, 3 tables | null | null | null | cs.RO cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ability to visually recognize objects is a fundamental skill for robotics
systems. Indeed, a large variety of tasks involving manipulation, navigation or
interaction with other agents, deeply depends on the accurate understanding of
the visual scene. Yet, at the time being, robots are lacking good visual
perceptual systems, which often become the main bottleneck preventing the use
of autonomous agents for real-world applications.
Lately in computer vision, systems that learn suitable visual representations
and based on multi-layer deep convolutional networks are showing remarkable
performance in tasks such as large-scale visual recognition and image
retrieval. To this regard, it is natural to ask whether such remarkable
performance would generalize also to the robotic setting.
In this paper we investigate such possibility, while taking further steps in
developing a computational vision system to be embedded on a robotic platform,
the iCub humanoid robot. In particular, we release a new dataset ({\sc
iCubWorld28}) that we use as a benchmark to address the question: {\it how many
objects can iCub recognize?} Our study is developed in a learning framework
which reflects the typical visual experience of a humanoid robot like the iCub.
Experiments shed interesting insights on the strength and weaknesses of current
computer vision approaches applied in real robotic settings.
| [
{
"version": "v1",
"created": "Mon, 13 Apr 2015 12:45:09 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Apr 2015 05:56:01 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Pasquale",
"Giulia",
""
],
[
"Ciliberto",
"Carlo",
""
],
[
"Odone",
"Francesca",
""
],
[
"Rosasco",
"Lorenzo",
""
],
[
"Natale",
"Lorenzo",
""
]
] | TITLE: Real-world Object Recognition with Off-the-shelf Deep Conv Nets: How
Many Objects can iCub Learn?
ABSTRACT: The ability to visually recognize objects is a fundamental skill for robotics
systems. Indeed, a large variety of tasks involving manipulation, navigation or
interaction with other agents, deeply depends on the accurate understanding of
the visual scene. Yet, at the time being, robots are lacking good visual
perceptual systems, which often become the main bottleneck preventing the use
of autonomous agents for real-world applications.
Lately in computer vision, systems that learn suitable visual representations
and based on multi-layer deep convolutional networks are showing remarkable
performance in tasks such as large-scale visual recognition and image
retrieval. To this regard, it is natural to ask whether such remarkable
performance would generalize also to the robotic setting.
In this paper we investigate such possibility, while taking further steps in
developing a computational vision system to be embedded on a robotic platform,
the iCub humanoid robot. In particular, we release a new dataset ({\sc
iCubWorld28}) that we use as a benchmark to address the question: {\it how many
objects can iCub recognize?} Our study is developed in a learning framework
which reflects the typical visual experience of a humanoid robot like the iCub.
Experiments shed interesting insights on the strength and weaknesses of current
computer vision approaches applied in real robotic settings.
| new_dataset | 0.960249 |
1504.03425 | Iftekhar Naim | Iftekhar Naim, M. Iftekhar Tanveer, Daniel Gildea, Mohammed (Ehsan)
Hoque | Automated Analysis and Prediction of Job Interview Performance | 14 pages, 8 figures, 6 tables | null | null | null | cs.HC cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a computational framework for automatically quantifying verbal and
nonverbal behaviors in the context of job interviews. The proposed framework is
trained by analyzing the videos of 138 interview sessions with 69
internship-seeking undergraduates at the Massachusetts Institute of Technology
(MIT). Our automated analysis includes facial expressions (e.g., smiles, head
gestures, facial tracking points), language (e.g., word counts, topic
modeling), and prosodic information (e.g., pitch, intonation, and pauses) of
the interviewees. The ground truth labels are derived by taking a weighted
average over the ratings of 9 independent judges. Our framework can
automatically predict the ratings for interview traits such as excitement,
friendliness, and engagement with correlation coefficients of 0.75 or higher,
and can quantify the relative importance of prosody, language, and facial
expressions. By analyzing the relative feature weights learned by the
regression models, our framework recommends to speak more fluently, use less
filler words, speak as "we" (vs. "I"), use more unique words, and smile more.
We also find that the students who were rated highly while answering the first
interview question were also rated highly overall (i.e., first impression
matters). Finally, our MIT Interview dataset will be made available to other
researchers to further validate and expand our findings.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 05:49:26 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Naim",
"Iftekhar",
"",
"Ehsan"
],
[
"Tanveer",
"M. Iftekhar",
"",
"Ehsan"
],
[
"Gildea",
"Daniel",
"",
"Ehsan"
],
[
"Mohammed",
"",
"",
"Ehsan"
],
[
"Hoque",
"",
""
]
] | TITLE: Automated Analysis and Prediction of Job Interview Performance
ABSTRACT: We present a computational framework for automatically quantifying verbal and
nonverbal behaviors in the context of job interviews. The proposed framework is
trained by analyzing the videos of 138 interview sessions with 69
internship-seeking undergraduates at the Massachusetts Institute of Technology
(MIT). Our automated analysis includes facial expressions (e.g., smiles, head
gestures, facial tracking points), language (e.g., word counts, topic
modeling), and prosodic information (e.g., pitch, intonation, and pauses) of
the interviewees. The ground truth labels are derived by taking a weighted
average over the ratings of 9 independent judges. Our framework can
automatically predict the ratings for interview traits such as excitement,
friendliness, and engagement with correlation coefficients of 0.75 or higher,
and can quantify the relative importance of prosody, language, and facial
expressions. By analyzing the relative feature weights learned by the
regression models, our framework recommends to speak more fluently, use less
filler words, speak as "we" (vs. "I"), use more unique words, and smile more.
We also find that the students who were rated highly while answering the first
interview question were also rated highly overall (i.e., first impression
matters). Finally, our MIT Interview dataset will be made available to other
researchers to further validate and expand our findings.
| new_dataset | 0.953622 |
1504.03486 | Mark Leake | Mark Leake | Analytical tools for single-molecule fluorescence imaging in cellulo | null | null | null | null | q-bio.BM physics.bio-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent technological advances in cutting-edge ultrasensitive fluorescence
microscopy have allowed single-molecule imaging experiments in living cells
across all three domains of life to become commonplace. Single-molecule
live-cell data is typically obtained in a low signal-to-noise ratio (SNR)
regime sometimes only marginally in excess of 1, in which a combination of
detector shot noise, sub-optimal probe photophysics, native cell
autofluorescence and intrinsically underlying stochastic of molecules result in
highly noisy datasets for which underlying true molecular behaviour is
non-trivial to discern. The ability to elucidate real molecular phenomena is
essential in relating experimental single-molecule observations to both the
biological system under study as well as offering insight into the fine details
of the physical and chemical environments of the living cell. To confront this
problem of faithful signal extraction and analysis in a noise-dominated regime,
the needle in a haystack challenge, such experiments benefit enormously from a
suite of objective, automated, high-throughput analysis tools that can home in
on the underlying molecular signature and generate meaningful statistics across
a large population of individual cells and molecules. Here, I discuss the
development and application of several analytical methods applied to real case
studies, including objective methods of segmenting cellular images from light
microscopy data, tools to robustly localize and track single
fluorescently-labelled molecules, algorithms to objectively interpret molecular
mobility, analysis protocols to reliably estimate molecular stoichiometry and
turnover, and methods to objectively render distributions of molecular
parameters
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 10:35:06 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Leake",
"Mark",
""
]
] | TITLE: Analytical tools for single-molecule fluorescence imaging in cellulo
ABSTRACT: Recent technological advances in cutting-edge ultrasensitive fluorescence
microscopy have allowed single-molecule imaging experiments in living cells
across all three domains of life to become commonplace. Single-molecule
live-cell data is typically obtained in a low signal-to-noise ratio (SNR)
regime sometimes only marginally in excess of 1, in which a combination of
detector shot noise, sub-optimal probe photophysics, native cell
autofluorescence and intrinsically underlying stochastic of molecules result in
highly noisy datasets for which underlying true molecular behaviour is
non-trivial to discern. The ability to elucidate real molecular phenomena is
essential in relating experimental single-molecule observations to both the
biological system under study as well as offering insight into the fine details
of the physical and chemical environments of the living cell. To confront this
problem of faithful signal extraction and analysis in a noise-dominated regime,
the needle in a haystack challenge, such experiments benefit enormously from a
suite of objective, automated, high-throughput analysis tools that can home in
on the underlying molecular signature and generate meaningful statistics across
a large population of individual cells and molecules. Here, I discuss the
development and application of several analytical methods applied to real case
studies, including objective methods of segmenting cellular images from light
microscopy data, tools to robustly localize and track single
fluorescently-labelled molecules, algorithms to objectively interpret molecular
mobility, analysis protocols to reliably estimate molecular stoichiometry and
turnover, and methods to objectively render distributions of molecular
parameters
| no_new_dataset | 0.948058 |
1504.03504 | Fang Wang | Fang Wang, Le Kang, Yi Li | Sketch-based 3D Shape Retrieval using Convolutional Neural Networks | CVPR 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Retrieving 3D models from 2D human sketches has received considerable
attention in the areas of graphics, image retrieval, and computer vision.
Almost always in state of the art approaches a large amount of "best views" are
computed for 3D models, with the hope that the query sketch matches one of
these 2D projections of 3D models using predefined features.
We argue that this two stage approach (view selection -- matching) is
pragmatic but also problematic because the "best views" are subjective and
ambiguous, which makes the matching inputs obscure. This imprecise nature of
matching further makes it challenging to choose features manually. Instead of
relying on the elusive concept of "best views" and the hand-crafted features,
we propose to define our views using a minimalism approach and learn features
for both sketches and views. Specifically, we drastically reduce the number of
views to only two predefined directions for the whole dataset. Then, we learn
two Siamese Convolutional Neural Networks (CNNs), one for the views and one for
the sketches. The loss function is defined on the within-domain as well as the
cross-domain similarities. Our experiments on three benchmark datasets
demonstrate that our method is significantly better than state of the art
approaches, and outperforms them in all conventional metrics.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 11:55:45 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Wang",
"Fang",
""
],
[
"Kang",
"Le",
""
],
[
"Li",
"Yi",
""
]
] | TITLE: Sketch-based 3D Shape Retrieval using Convolutional Neural Networks
ABSTRACT: Retrieving 3D models from 2D human sketches has received considerable
attention in the areas of graphics, image retrieval, and computer vision.
Almost always in state of the art approaches a large amount of "best views" are
computed for 3D models, with the hope that the query sketch matches one of
these 2D projections of 3D models using predefined features.
We argue that this two stage approach (view selection -- matching) is
pragmatic but also problematic because the "best views" are subjective and
ambiguous, which makes the matching inputs obscure. This imprecise nature of
matching further makes it challenging to choose features manually. Instead of
relying on the elusive concept of "best views" and the hand-crafted features,
we propose to define our views using a minimalism approach and learn features
for both sketches and views. Specifically, we drastically reduce the number of
views to only two predefined directions for the whole dataset. Then, we learn
two Siamese Convolutional Neural Networks (CNNs), one for the views and one for
the sketches. The loss function is defined on the within-domain as well as the
cross-domain similarities. Our experiments on three benchmark datasets
demonstrate that our method is significantly better than state of the art
approaches, and outperforms them in all conventional metrics.
| no_new_dataset | 0.949012 |
1504.03522 | Lukas Neumann | Luk\'a\v{s} Neumann, Ji\v{r}\'i Matas | Efficient Scene Text Localization and Recognition with Local Character
Refinement | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | An unconstrained end-to-end text localization and recognition method is
presented. The method detects initial text hypothesis in a single pass by an
efficient region-based method and subsequently refines the text hypothesis
using a more robust local text model, which deviates from the common assumption
of region-based methods that all characters are detected as connected
components.
Additionally, a novel feature based on character stroke area estimation is
introduced. The feature is efficiently computed from a region distance map, it
is invariant to scaling and rotations and allows to efficiently detect text
regions regardless of what portion of text they capture.
The method runs in real time and achieves state-of-the-art text localization
and recognition results on the ICDAR 2013 Robust Reading dataset.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 12:42:56 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Neumann",
"Lukáš",
""
],
[
"Matas",
"Jiří",
""
]
] | TITLE: Efficient Scene Text Localization and Recognition with Local Character
Refinement
ABSTRACT: An unconstrained end-to-end text localization and recognition method is
presented. The method detects initial text hypothesis in a single pass by an
efficient region-based method and subsequently refines the text hypothesis
using a more robust local text model, which deviates from the common assumption
of region-based methods that all characters are detected as connected
components.
Additionally, a novel feature based on character stroke area estimation is
introduced. The feature is efficiently computed from a region distance map, it
is invariant to scaling and rotations and allows to efficiently detect text
regions regardless of what portion of text they capture.
The method runs in real time and achieves state-of-the-art text localization
and recognition results on the ICDAR 2013 Robust Reading dataset.
| no_new_dataset | 0.94868 |
1504.03659 | Azad Dehghan Mr | Azad Dehghan | Temporal ordering of clinical events | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This report describes a minimalistic set of methods engineered to anchor
clinical events onto a temporal space. Specifically, we describe methods to
extract clinical events (e.g., Problems, Treatments and Tests), temporal
expressions (i.e., time, date, duration, and frequency), and temporal links
(e.g., Before, After, Overlap) between events and temporal entities. These
methods are developed and validated using high quality datasets.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 18:48:58 GMT"
}
] | 2015-04-15T00:00:00 | [
[
"Dehghan",
"Azad",
""
]
] | TITLE: Temporal ordering of clinical events
ABSTRACT: This report describes a minimalistic set of methods engineered to anchor
clinical events onto a temporal space. Specifically, we describe methods to
extract clinical events (e.g., Problems, Treatments and Tests), temporal
expressions (i.e., time, date, duration, and frequency), and temporal links
(e.g., Before, After, Overlap) between events and temporal entities. These
methods are developed and validated using high quality datasets.
| no_new_dataset | 0.946448 |
1412.6597 | Tom Paine | Tom Le Paine, Pooya Khorrami, Wei Han, Thomas S. Huang | An Analysis of Unsupervised Pre-training in Light of Recent Advances | Accepted as a workshop contribution to ICLR 2015 | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Convolutional neural networks perform well on object recognition because of a
number of recent advances: rectified linear units (ReLUs), data augmentation,
dropout, and large labelled datasets. Unsupervised data has been proposed as
another way to improve performance. Unfortunately, unsupervised pre-training is
not used by state-of-the-art methods leading to the following question: Is
unsupervised pre-training still useful given recent advances? If so, when? We
answer this in three parts: we 1) develop an unsupervised method that
incorporates ReLUs and recent unsupervised regularization techniques, 2)
analyze the benefits of unsupervised pre-training compared to data augmentation
and dropout on CIFAR-10 while varying the ratio of unsupervised to supervised
samples, 3) verify our findings on STL-10. We discover unsupervised
pre-training, as expected, helps when the ratio of unsupervised to supervised
samples is high, and surprisingly, hurts when the ratio is low. We also use
unsupervised pre-training with additional color augmentation to achieve near
state-of-the-art performance on STL-10.
| [
{
"version": "v1",
"created": "Sat, 20 Dec 2014 04:20:55 GMT"
},
{
"version": "v2",
"created": "Tue, 27 Jan 2015 22:03:40 GMT"
},
{
"version": "v3",
"created": "Mon, 2 Mar 2015 21:05:34 GMT"
},
{
"version": "v4",
"created": "Fri, 10 Apr 2015 21:26:31 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Paine",
"Tom Le",
""
],
[
"Khorrami",
"Pooya",
""
],
[
"Han",
"Wei",
""
],
[
"Huang",
"Thomas S.",
""
]
] | TITLE: An Analysis of Unsupervised Pre-training in Light of Recent Advances
ABSTRACT: Convolutional neural networks perform well on object recognition because of a
number of recent advances: rectified linear units (ReLUs), data augmentation,
dropout, and large labelled datasets. Unsupervised data has been proposed as
another way to improve performance. Unfortunately, unsupervised pre-training is
not used by state-of-the-art methods leading to the following question: Is
unsupervised pre-training still useful given recent advances? If so, when? We
answer this in three parts: we 1) develop an unsupervised method that
incorporates ReLUs and recent unsupervised regularization techniques, 2)
analyze the benefits of unsupervised pre-training compared to data augmentation
and dropout on CIFAR-10 while varying the ratio of unsupervised to supervised
samples, 3) verify our findings on STL-10. We discover unsupervised
pre-training, as expected, helps when the ratio of unsupervised to supervised
samples is high, and surprisingly, hurts when the ratio is low. We also use
unsupervised pre-training with additional color augmentation to achieve near
state-of-the-art performance on STL-10.
| no_new_dataset | 0.945399 |
1412.6806 | Alexey Dosovitskiy | Jost Tobias Springenberg, Alexey Dosovitskiy, Thomas Brox, Martin
Riedmiller | Striving for Simplicity: The All Convolutional Net | accepted to ICLR-2015 workshop track; no changes other than style | null | null | null | cs.LG cs.CV cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most modern convolutional neural networks (CNNs) used for object recognition
are built using the same principles: Alternating convolution and max-pooling
layers followed by a small number of fully connected layers. We re-evaluate the
state of the art for object recognition from small images with convolutional
networks, questioning the necessity of different components in the pipeline. We
find that max-pooling can simply be replaced by a convolutional layer with
increased stride without loss in accuracy on several image recognition
benchmarks. Following this finding -- and building on other recent work for
finding simple network structures -- we propose a new architecture that
consists solely of convolutional layers and yields competitive or state of the
art performance on several object recognition datasets (CIFAR-10, CIFAR-100,
ImageNet). To analyze the network we introduce a new variant of the
"deconvolution approach" for visualizing features learned by CNNs, which can be
applied to a broader range of network structures than existing approaches.
| [
{
"version": "v1",
"created": "Sun, 21 Dec 2014 16:16:37 GMT"
},
{
"version": "v2",
"created": "Mon, 2 Mar 2015 21:44:06 GMT"
},
{
"version": "v3",
"created": "Mon, 13 Apr 2015 07:58:17 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Springenberg",
"Jost Tobias",
""
],
[
"Dosovitskiy",
"Alexey",
""
],
[
"Brox",
"Thomas",
""
],
[
"Riedmiller",
"Martin",
""
]
] | TITLE: Striving for Simplicity: The All Convolutional Net
ABSTRACT: Most modern convolutional neural networks (CNNs) used for object recognition
are built using the same principles: Alternating convolution and max-pooling
layers followed by a small number of fully connected layers. We re-evaluate the
state of the art for object recognition from small images with convolutional
networks, questioning the necessity of different components in the pipeline. We
find that max-pooling can simply be replaced by a convolutional layer with
increased stride without loss in accuracy on several image recognition
benchmarks. Following this finding -- and building on other recent work for
finding simple network structures -- we propose a new architecture that
consists solely of convolutional layers and yields competitive or state of the
art performance on several object recognition datasets (CIFAR-10, CIFAR-100,
ImageNet). To analyze the network we introduce a new variant of the
"deconvolution approach" for visualizing features learned by CNNs, which can be
applied to a broader range of network structures than existing approaches.
| no_new_dataset | 0.949529 |
1412.8419 | Pedro O. Pinheiro | Remi Lebret and Pedro O. Pinheiro and Ronan Collobert | Simple Image Description Generator via a Linear Phrase-Based Approach | Accepted as a workshop paper at ICLR 2015 | null | null | null | cs.CL cs.CV cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generating a novel textual description of an image is an interesting problem
that connects computer vision and natural language processing. In this paper,
we present a simple model that is able to generate descriptive sentences given
a sample image. This model has a strong focus on the syntax of the
descriptions. We train a purely bilinear model that learns a metric between an
image representation (generated from a previously trained Convolutional Neural
Network) and phrases that are used to described them. The system is then able
to infer phrases from a given image sample. Based on caption syntax statistics,
we propose a simple language model that can produce relevant descriptions for a
given test image using the phrases inferred. Our approach, which is
considerably simpler than state-of-the-art models, achieves comparable results
on the recently release Microsoft COCO dataset.
| [
{
"version": "v1",
"created": "Mon, 29 Dec 2014 18:43:10 GMT"
},
{
"version": "v2",
"created": "Wed, 18 Mar 2015 05:09:13 GMT"
},
{
"version": "v3",
"created": "Sat, 11 Apr 2015 03:53:26 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Lebret",
"Remi",
""
],
[
"Pinheiro",
"Pedro O.",
""
],
[
"Collobert",
"Ronan",
""
]
] | TITLE: Simple Image Description Generator via a Linear Phrase-Based Approach
ABSTRACT: Generating a novel textual description of an image is an interesting problem
that connects computer vision and natural language processing. In this paper,
we present a simple model that is able to generate descriptive sentences given
a sample image. This model has a strong focus on the syntax of the
descriptions. We train a purely bilinear model that learns a metric between an
image representation (generated from a previously trained Convolutional Neural
Network) and phrases that are used to described them. The system is then able
to infer phrases from a given image sample. Based on caption syntax statistics,
we propose a simple language model that can produce relevant descriptions for a
given test image using the phrases inferred. Our approach, which is
considerably simpler than state-of-the-art models, achieves comparable results
on the recently release Microsoft COCO dataset.
| no_new_dataset | 0.944587 |
1502.07645 | Yu-Xiang Wang | Yu-Xiang Wang, Stephen E. Fienberg, Alex Smola | Privacy for Free: Posterior Sampling and Stochastic Gradient Monte Carlo | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the problem of Bayesian learning on sensitive datasets and
present two simple but somewhat surprising results that connect Bayesian
learning to "differential privacy:, a cryptographic approach to protect
individual-level privacy while permiting database-level utility. Specifically,
we show that that under standard assumptions, getting one single sample from a
posterior distribution is differentially private "for free". We will see that
estimator is statistically consistent, near optimal and computationally
tractable whenever the Bayesian model of interest is consistent, optimal and
tractable. Similarly but separately, we show that a recent line of works that
use stochastic gradient for Hybrid Monte Carlo (HMC) sampling also preserve
differentially privacy with minor or no modifications of the algorithmic
procedure at all, these observations lead to an "anytime" algorithm for
Bayesian learning under privacy constraint. We demonstrate that it performs
much better than the state-of-the-art differential private methods on synthetic
and real datasets.
| [
{
"version": "v1",
"created": "Thu, 26 Feb 2015 17:38:47 GMT"
},
{
"version": "v2",
"created": "Sun, 12 Apr 2015 02:53:05 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Wang",
"Yu-Xiang",
""
],
[
"Fienberg",
"Stephen E.",
""
],
[
"Smola",
"Alex",
""
]
] | TITLE: Privacy for Free: Posterior Sampling and Stochastic Gradient Monte Carlo
ABSTRACT: We consider the problem of Bayesian learning on sensitive datasets and
present two simple but somewhat surprising results that connect Bayesian
learning to "differential privacy:, a cryptographic approach to protect
individual-level privacy while permiting database-level utility. Specifically,
we show that that under standard assumptions, getting one single sample from a
posterior distribution is differentially private "for free". We will see that
estimator is statistically consistent, near optimal and computationally
tractable whenever the Bayesian model of interest is consistent, optimal and
tractable. Similarly but separately, we show that a recent line of works that
use stochastic gradient for Hybrid Monte Carlo (HMC) sampling also preserve
differentially privacy with minor or no modifications of the algorithmic
procedure at all, these observations lead to an "anytime" algorithm for
Bayesian learning under privacy constraint. We demonstrate that it performs
much better than the state-of-the-art differential private methods on synthetic
and real datasets.
| no_new_dataset | 0.947914 |
1503.02801 | Jiaming Xu | Jiaming Xu, Bo Xu, Guanhua Tian, Jun Zhao, Fangyuan Wang, Hongwei Hao | Short Text Hashing Improved by Integrating Multi-Granularity Topics and
Tags | 12 pages, accepted at CICLing 2015 | null | 10.1007/978-3-319-18111-0_33 | null | cs.IR cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Due to computational and storage efficiencies of compact binary codes,
hashing has been widely used for large-scale similarity search. Unfortunately,
many existing hashing methods based on observed keyword features are not
effective for short texts due to the sparseness and shortness. Recently, some
researchers try to utilize latent topics of certain granularity to preserve
semantic similarity in hash codes beyond keyword matching. However, topics of
certain granularity are not adequate to represent the intrinsic semantic
information. In this paper, we present a novel unified approach for short text
Hashing using Multi-granularity Topics and Tags, dubbed HMTT. In particular, we
propose a selection method to choose the optimal multi-granularity topics
depending on the type of dataset, and design two distinct hashing strategies to
incorporate multi-granularity topics. We also propose a simple and effective
method to exploit tags to enhance the similarity of related texts. We carry out
extensive experiments on one short text dataset as well as on one normal text
dataset. The results demonstrate that our approach is effective and
significantly outperforms baselines on several evaluation metrics.
| [
{
"version": "v1",
"created": "Tue, 10 Mar 2015 07:51:59 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Xu",
"Jiaming",
""
],
[
"Xu",
"Bo",
""
],
[
"Tian",
"Guanhua",
""
],
[
"Zhao",
"Jun",
""
],
[
"Wang",
"Fangyuan",
""
],
[
"Hao",
"Hongwei",
""
]
] | TITLE: Short Text Hashing Improved by Integrating Multi-Granularity Topics and
Tags
ABSTRACT: Due to computational and storage efficiencies of compact binary codes,
hashing has been widely used for large-scale similarity search. Unfortunately,
many existing hashing methods based on observed keyword features are not
effective for short texts due to the sparseness and shortness. Recently, some
researchers try to utilize latent topics of certain granularity to preserve
semantic similarity in hash codes beyond keyword matching. However, topics of
certain granularity are not adequate to represent the intrinsic semantic
information. In this paper, we present a novel unified approach for short text
Hashing using Multi-granularity Topics and Tags, dubbed HMTT. In particular, we
propose a selection method to choose the optimal multi-granularity topics
depending on the type of dataset, and design two distinct hashing strategies to
incorporate multi-granularity topics. We also propose a simple and effective
method to exploit tags to enhance the similarity of related texts. We carry out
extensive experiments on one short text dataset as well as on one normal text
dataset. The results demonstrate that our approach is effective and
significantly outperforms baselines on several evaluation metrics.
| no_new_dataset | 0.944893 |
1503.06813 | Tarek El-Gaaly | Haopeng Zhang, Tarek El-Gaaly, Ahmed Elgammal, Zhiguo Jiang | Factorization of View-Object Manifolds for Joint Object Recognition and
Pose Estimation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Due to large variations in shape, appearance, and viewing conditions, object
recognition is a key precursory challenge in the fields of object manipulation
and robotic/AI visual reasoning in general. Recognizing object categories,
particular instances of objects and viewpoints/poses of objects are three
critical subproblems robots must solve in order to accurately grasp/manipulate
objects and reason about their environments. Multi-view images of the same
object lie on intrinsic low-dimensional manifolds in descriptor spaces (e.g.
visual/depth descriptor spaces). These object manifolds share the same topology
despite being geometrically different. Each object manifold can be represented
as a deformed version of a unified manifold. The object manifolds can thus be
parameterized by its homeomorphic mapping/reconstruction from the unified
manifold. In this work, we develop a novel framework to jointly solve the three
challenging recognition sub-problems, by explicitly modeling the deformations
of object manifolds and factorizing it in a view-invariant space for
recognition. We perform extensive experiments on several challenging datasets
and achieve state-of-the-art results.
| [
{
"version": "v1",
"created": "Mon, 23 Mar 2015 20:05:36 GMT"
},
{
"version": "v2",
"created": "Mon, 13 Apr 2015 02:59:41 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Zhang",
"Haopeng",
""
],
[
"El-Gaaly",
"Tarek",
""
],
[
"Elgammal",
"Ahmed",
""
],
[
"Jiang",
"Zhiguo",
""
]
] | TITLE: Factorization of View-Object Manifolds for Joint Object Recognition and
Pose Estimation
ABSTRACT: Due to large variations in shape, appearance, and viewing conditions, object
recognition is a key precursory challenge in the fields of object manipulation
and robotic/AI visual reasoning in general. Recognizing object categories,
particular instances of objects and viewpoints/poses of objects are three
critical subproblems robots must solve in order to accurately grasp/manipulate
objects and reason about their environments. Multi-view images of the same
object lie on intrinsic low-dimensional manifolds in descriptor spaces (e.g.
visual/depth descriptor spaces). These object manifolds share the same topology
despite being geometrically different. Each object manifold can be represented
as a deformed version of a unified manifold. The object manifolds can thus be
parameterized by its homeomorphic mapping/reconstruction from the unified
manifold. In this work, we develop a novel framework to jointly solve the three
challenging recognition sub-problems, by explicitly modeling the deformations
of object manifolds and factorizing it in a view-invariant space for
recognition. We perform extensive experiments on several challenging datasets
and achieve state-of-the-art results.
| no_new_dataset | 0.948917 |
1503.08909 | George Toderici | Joe Yue-Hei Ng and Matthew Hausknecht and Sudheendra Vijayanarasimhan
and Oriol Vinyals and Rajat Monga and George Toderici | Beyond Short Snippets: Deep Networks for Video Classification | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Convolutional neural networks (CNNs) have been extensively applied for image
recognition problems giving state-of-the-art results on recognition, detection,
segmentation and retrieval. In this work we propose and evaluate several deep
neural network architectures to combine image information across a video over
longer time periods than previously attempted. We propose two methods capable
of handling full length videos. The first method explores various convolutional
temporal feature pooling architectures, examining the various design choices
which need to be made when adapting a CNN for this task. The second proposed
method explicitly models the video as an ordered sequence of frames. For this
purpose we employ a recurrent neural network that uses Long Short-Term Memory
(LSTM) cells which are connected to the output of the underlying CNN. Our best
networks exhibit significant performance improvements over previously published
results on the Sports 1 million dataset (73.1% vs. 60.9%) and the UCF-101
datasets with (88.6% vs. 88.0%) and without additional optical flow information
(82.6% vs. 72.8%).
| [
{
"version": "v1",
"created": "Tue, 31 Mar 2015 04:34:12 GMT"
},
{
"version": "v2",
"created": "Mon, 13 Apr 2015 19:44:25 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Ng",
"Joe Yue-Hei",
""
],
[
"Hausknecht",
"Matthew",
""
],
[
"Vijayanarasimhan",
"Sudheendra",
""
],
[
"Vinyals",
"Oriol",
""
],
[
"Monga",
"Rajat",
""
],
[
"Toderici",
"George",
""
]
] | TITLE: Beyond Short Snippets: Deep Networks for Video Classification
ABSTRACT: Convolutional neural networks (CNNs) have been extensively applied for image
recognition problems giving state-of-the-art results on recognition, detection,
segmentation and retrieval. In this work we propose and evaluate several deep
neural network architectures to combine image information across a video over
longer time periods than previously attempted. We propose two methods capable
of handling full length videos. The first method explores various convolutional
temporal feature pooling architectures, examining the various design choices
which need to be made when adapting a CNN for this task. The second proposed
method explicitly models the video as an ordered sequence of frames. For this
purpose we employ a recurrent neural network that uses Long Short-Term Memory
(LSTM) cells which are connected to the output of the underlying CNN. Our best
networks exhibit significant performance improvements over previously published
results on the Sports 1 million dataset (73.1% vs. 60.9%) and the UCF-101
datasets with (88.6% vs. 88.0%) and without additional optical flow information
(82.6% vs. 72.8%).
| no_new_dataset | 0.953622 |
1504.02902 | Alexander Kalmanovich | Alexander Kalmanovich and Gal Chechik | Gradual Training Method for Denoising Auto Encoders | arXiv admin note: substantial text overlap with arXiv:1412.6257 | 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": "Sat, 11 Apr 2015 17:51:41 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Kalmanovich",
"Alexander",
""
],
[
"Chechik",
"Gal",
""
]
] | TITLE: Gradual Training Method for 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.
| no_new_dataset | 0.949482 |
1504.02975 | F. Ozgur Catak | Ferhat \"Ozg\"ur \c{C}atak | Classification with Extreme Learning Machine and Ensemble Algorithms
Over Randomly Partitioned Data | In Turkish, SIU | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this age of Big Data, machine learning based data mining methods are
extensively used to inspect large scale data sets. Deriving applicable
predictive modeling from these type of data sets is a challenging obstacle
because of their high complexity. Opportunity with high data availability
levels, automated classification of data sets has become a critical and
complicated function. In this paper, the power of applying MapReduce based
Distributed AdaBoosting of Extreme Learning Machine (ELM) are explored to build
reliable predictive bag of classification models. Thus, (i) dataset ensembles
are build; (ii) ELM algorithm is used to build weak classification models; and
(iii) build a strong classification model from a set of weak classification
models. This training model is applied to the publicly available knowledge
discovery and data mining datasets.
| [
{
"version": "v1",
"created": "Sun, 12 Apr 2015 14:03:25 GMT"
}
] | 2015-04-14T00:00:00 | [
[
"Çatak",
"Ferhat Özgür",
""
]
] | TITLE: Classification with Extreme Learning Machine and Ensemble Algorithms
Over Randomly Partitioned Data
ABSTRACT: In this age of Big Data, machine learning based data mining methods are
extensively used to inspect large scale data sets. Deriving applicable
predictive modeling from these type of data sets is a challenging obstacle
because of their high complexity. Opportunity with high data availability
levels, automated classification of data sets has become a critical and
complicated function. In this paper, the power of applying MapReduce based
Distributed AdaBoosting of Extreme Learning Machine (ELM) are explored to build
reliable predictive bag of classification models. Thus, (i) dataset ensembles
are build; (ii) ELM algorithm is used to build weak classification models; and
(iii) build a strong classification model from a set of weak classification
models. This training model is applied to the publicly available knowledge
discovery and data mining datasets.
| no_new_dataset | 0.94545 |
1301.1275 | Luca Allodi | Luca Allodi, Fabio Massacci | My Software has a Vulnerability, should I worry? | 12 pages, 4 figures | ACM TISSEC Vol 17 Issue 1, 2014 | 10.1145/2630069 | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | (U.S) Rule-based policies to mitigate software risk suggest to use the CVSS
score to measure the individual vulnerability risk and act accordingly: an HIGH
CVSS score according to the NVD (National (U.S.) Vulnerability Database) is
therefore translated into a "Yes". A key issue is whether such rule is
economically sensible, in particular if reported vulnerabilities have been
actually exploited in the wild, and whether the risk score do actually match
the risk of actual exploitation.
We compare the NVD dataset with two additional datasets, the EDB for the
white market of vulnerabilities (such as those present in Metasploit), and the
EKITS for the exploits traded in the black market. We benchmark them against
Symantec's threat explorer dataset (SYM) of actual exploit in the wild. We
analyze the whole spectrum of CVSS submetrics and use these characteristics to
perform a case-controlled analysis of CVSS scores (similar to those used to
link lung cancer and smoking) to test its reliability as a risk factor for
actual exploitation.
We conclude that (a) fixing just because a high CVSS score in NVD only yields
negligible risk reduction, (b) the additional existence of proof of concepts
exploits (e.g. in EDB) may yield some additional but not large risk reduction,
(c) fixing in response to presence in black markets yields the equivalent risk
reduction of wearing safety belt in cars (you might also die but still..). On
the negative side, our study shows that as industry we miss a metric with high
specificity (ruling out vulns for which we shouldn't worry).
In order to address the feedback from BlackHat 2013's audience, the final
revision (V3) provides additional data in Appendix A detailing how the control
variables in the study affect the results.
| [
{
"version": "v1",
"created": "Mon, 7 Jan 2013 17:32:36 GMT"
},
{
"version": "v2",
"created": "Mon, 5 Aug 2013 21:09:41 GMT"
},
{
"version": "v3",
"created": "Tue, 24 Sep 2013 14:05:27 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Allodi",
"Luca",
""
],
[
"Massacci",
"Fabio",
""
]
] | TITLE: My Software has a Vulnerability, should I worry?
ABSTRACT: (U.S) Rule-based policies to mitigate software risk suggest to use the CVSS
score to measure the individual vulnerability risk and act accordingly: an HIGH
CVSS score according to the NVD (National (U.S.) Vulnerability Database) is
therefore translated into a "Yes". A key issue is whether such rule is
economically sensible, in particular if reported vulnerabilities have been
actually exploited in the wild, and whether the risk score do actually match
the risk of actual exploitation.
We compare the NVD dataset with two additional datasets, the EDB for the
white market of vulnerabilities (such as those present in Metasploit), and the
EKITS for the exploits traded in the black market. We benchmark them against
Symantec's threat explorer dataset (SYM) of actual exploit in the wild. We
analyze the whole spectrum of CVSS submetrics and use these characteristics to
perform a case-controlled analysis of CVSS scores (similar to those used to
link lung cancer and smoking) to test its reliability as a risk factor for
actual exploitation.
We conclude that (a) fixing just because a high CVSS score in NVD only yields
negligible risk reduction, (b) the additional existence of proof of concepts
exploits (e.g. in EDB) may yield some additional but not large risk reduction,
(c) fixing in response to presence in black markets yields the equivalent risk
reduction of wearing safety belt in cars (you might also die but still..). On
the negative side, our study shows that as industry we miss a metric with high
specificity (ruling out vulns for which we shouldn't worry).
In order to address the feedback from BlackHat 2013's audience, the final
revision (V3) provides additional data in Appendix A detailing how the control
variables in the study affect the results.
| no_new_dataset | 0.902995 |
1406.2080 | Sainbayar Sukhbaatar | Sainbayar Sukhbaatar, Joan Bruna, Manohar Paluri, Lubomir Bourdev and
Rob Fergus | Training Convolutional Networks with Noisy Labels | Accepted as a workshop contribution at ICLR 2015 | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The availability of large labeled datasets has allowed Convolutional Network
models to achieve impressive recognition results. However, in many settings
manual annotation of the data is impractical; instead our data has noisy
labels, i.e. there is some freely available label for each image which may or
may not be accurate. In this paper, we explore the performance of
discriminatively-trained Convnets when trained on such noisy data. We introduce
an extra noise layer into the network which adapts the network outputs to match
the noisy label distribution. The parameters of this noise layer can be
estimated as part of the training process and involve simple modifications to
current training infrastructures for deep networks. We demonstrate the
approaches on several datasets, including large scale experiments on the
ImageNet classification benchmark.
| [
{
"version": "v1",
"created": "Mon, 9 Jun 2014 05:45:12 GMT"
},
{
"version": "v2",
"created": "Sat, 20 Dec 2014 21:10:03 GMT"
},
{
"version": "v3",
"created": "Tue, 3 Mar 2015 21:13:47 GMT"
},
{
"version": "v4",
"created": "Fri, 10 Apr 2015 16:44:00 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Sukhbaatar",
"Sainbayar",
""
],
[
"Bruna",
"Joan",
""
],
[
"Paluri",
"Manohar",
""
],
[
"Bourdev",
"Lubomir",
""
],
[
"Fergus",
"Rob",
""
]
] | TITLE: Training Convolutional Networks with Noisy Labels
ABSTRACT: The availability of large labeled datasets has allowed Convolutional Network
models to achieve impressive recognition results. However, in many settings
manual annotation of the data is impractical; instead our data has noisy
labels, i.e. there is some freely available label for each image which may or
may not be accurate. In this paper, we explore the performance of
discriminatively-trained Convnets when trained on such noisy data. We introduce
an extra noise layer into the network which adapts the network outputs to match
the noisy label distribution. The parameters of this noise layer can be
estimated as part of the training process and involve simple modifications to
current training infrastructures for deep networks. We demonstrate the
approaches on several datasets, including large scale experiments on the
ImageNet classification benchmark.
| no_new_dataset | 0.948822 |
1409.1556 | Karen Simonyan | Karen Simonyan, Andrew Zisserman | Very Deep Convolutional Networks for Large-Scale Image Recognition | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we investigate the effect of the convolutional network depth on
its accuracy in the large-scale image recognition setting. Our main
contribution is a thorough evaluation of networks of increasing depth using an
architecture with very small (3x3) convolution filters, which shows that a
significant improvement on the prior-art configurations can be achieved by
pushing the depth to 16-19 weight layers. These findings were the basis of our
ImageNet Challenge 2014 submission, where our team secured the first and the
second places in the localisation and classification tracks respectively. We
also show that our representations generalise well to other datasets, where
they achieve state-of-the-art results. We have made our two best-performing
ConvNet models publicly available to facilitate further research on the use of
deep visual representations in computer vision.
| [
{
"version": "v1",
"created": "Thu, 4 Sep 2014 19:48:04 GMT"
},
{
"version": "v2",
"created": "Mon, 15 Sep 2014 19:58:29 GMT"
},
{
"version": "v3",
"created": "Tue, 18 Nov 2014 20:43:11 GMT"
},
{
"version": "v4",
"created": "Fri, 19 Dec 2014 20:01:21 GMT"
},
{
"version": "v5",
"created": "Tue, 23 Dec 2014 20:05:00 GMT"
},
{
"version": "v6",
"created": "Fri, 10 Apr 2015 16:25:04 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Simonyan",
"Karen",
""
],
[
"Zisserman",
"Andrew",
""
]
] | TITLE: Very Deep Convolutional Networks for Large-Scale Image Recognition
ABSTRACT: In this work we investigate the effect of the convolutional network depth on
its accuracy in the large-scale image recognition setting. Our main
contribution is a thorough evaluation of networks of increasing depth using an
architecture with very small (3x3) convolution filters, which shows that a
significant improvement on the prior-art configurations can be achieved by
pushing the depth to 16-19 weight layers. These findings were the basis of our
ImageNet Challenge 2014 submission, where our team secured the first and the
second places in the localisation and classification tracks respectively. We
also show that our representations generalise well to other datasets, where
they achieve state-of-the-art results. We have made our two best-performing
ConvNet models publicly available to facilitate further research on the use of
deep visual representations in computer vision.
| no_new_dataset | 0.951639 |
1410.1165 | Rupesh Kumar Srivastava | Rupesh Kumar Srivastava, Jonathan Masci, Faustino Gomez, J\"urgen
Schmidhuber | Understanding Locally Competitive Networks | 9 pages + 2 supplementary, Accepted to ICLR 2015 Conference track | null | null | null | cs.NE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently proposed neural network activation functions such as rectified
linear, maxout, and local winner-take-all have allowed for faster and more
effective training of deep neural architectures on large and complex datasets.
The common trait among these functions is that they implement local competition
between small groups of computational units within a layer, so that only part
of the network is activated for any given input pattern. In this paper, we
attempt to visualize and understand this self-modularization, and suggest a
unified explanation for the beneficial properties of such networks. We also
show how our insights can be directly useful for efficiently performing
retrieval over large datasets using neural networks.
| [
{
"version": "v1",
"created": "Sun, 5 Oct 2014 14:46:47 GMT"
},
{
"version": "v2",
"created": "Mon, 22 Dec 2014 20:07:17 GMT"
},
{
"version": "v3",
"created": "Thu, 9 Apr 2015 01:22:49 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Srivastava",
"Rupesh Kumar",
""
],
[
"Masci",
"Jonathan",
""
],
[
"Gomez",
"Faustino",
""
],
[
"Schmidhuber",
"Jürgen",
""
]
] | TITLE: Understanding Locally Competitive Networks
ABSTRACT: Recently proposed neural network activation functions such as rectified
linear, maxout, and local winner-take-all have allowed for faster and more
effective training of deep neural architectures on large and complex datasets.
The common trait among these functions is that they implement local competition
between small groups of computational units within a layer, so that only part
of the network is activated for any given input pattern. In this paper, we
attempt to visualize and understand this self-modularization, and suggest a
unified explanation for the beneficial properties of such networks. We also
show how our insights can be directly useful for efficiently performing
retrieval over large datasets using neural networks.
| no_new_dataset | 0.94801 |
1410.8516 | Laurent Dinh | Laurent Dinh, David Krueger and Yoshua Bengio | NICE: Non-linear Independent Components Estimation | 11 pages and 2 pages Appendix, workshop paper at ICLR 2015 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a deep learning framework for modeling complex high-dimensional
densities called Non-linear Independent Component Estimation (NICE). It is
based on the idea that a good representation is one in which the data has a
distribution that is easy to model. For this purpose, a non-linear
deterministic transformation of the data is learned that maps it to a latent
space so as to make the transformed data conform to a factorized distribution,
i.e., resulting in independent latent variables. We parametrize this
transformation so that computing the Jacobian determinant and inverse transform
is trivial, yet we maintain the ability to learn complex non-linear
transformations, via a composition of simple building blocks, each based on a
deep neural network. The training criterion is simply the exact log-likelihood,
which is tractable. Unbiased ancestral sampling is also easy. We show that this
approach yields good generative models on four image datasets and can be used
for inpainting.
| [
{
"version": "v1",
"created": "Thu, 30 Oct 2014 19:44:20 GMT"
},
{
"version": "v2",
"created": "Fri, 19 Dec 2014 22:40:18 GMT"
},
{
"version": "v3",
"created": "Tue, 6 Jan 2015 18:10:44 GMT"
},
{
"version": "v4",
"created": "Mon, 9 Mar 2015 18:06:58 GMT"
},
{
"version": "v5",
"created": "Thu, 12 Mar 2015 06:25:20 GMT"
},
{
"version": "v6",
"created": "Fri, 10 Apr 2015 12:27:56 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Dinh",
"Laurent",
""
],
[
"Krueger",
"David",
""
],
[
"Bengio",
"Yoshua",
""
]
] | TITLE: NICE: Non-linear Independent Components Estimation
ABSTRACT: We propose a deep learning framework for modeling complex high-dimensional
densities called Non-linear Independent Component Estimation (NICE). It is
based on the idea that a good representation is one in which the data has a
distribution that is easy to model. For this purpose, a non-linear
deterministic transformation of the data is learned that maps it to a latent
space so as to make the transformed data conform to a factorized distribution,
i.e., resulting in independent latent variables. We parametrize this
transformation so that computing the Jacobian determinant and inverse transform
is trivial, yet we maintain the ability to learn complex non-linear
transformations, via a composition of simple building blocks, each based on a
deep neural network. The training criterion is simply the exact log-likelihood,
which is tractable. Unbiased ancestral sampling is also easy. We show that this
approach yields good generative models on four image datasets and can be used
for inpainting.
| no_new_dataset | 0.944638 |
1503.08663 | Guanbin Li | Guanbin Li and Yizhou Yu | Visual Saliency Based on Multiscale Deep Features | To appear in CVPR 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Visual saliency is a fundamental problem in both cognitive and computational
sciences, including computer vision. In this CVPR 2015 paper, we discover that
a high-quality visual saliency model can be trained with multiscale features
extracted using a popular deep learning architecture, convolutional neural
networks (CNNs), which have had many successes in visual recognition tasks. For
learning such saliency models, we introduce a neural network architecture,
which has fully connected layers on top of CNNs responsible for extracting
features at three different scales. We then propose a refinement method to
enhance the spatial coherence of our saliency results. Finally, aggregating
multiple saliency maps computed for different levels of image segmentation can
further boost the performance, yielding saliency maps better than those
generated from a single segmentation. To promote further research and
evaluation of visual saliency models, we also construct a new large database of
4447 challenging images and their pixelwise saliency annotation. Experimental
results demonstrate that our proposed method is capable of achieving
state-of-the-art performance on all public benchmarks, improving the F-Measure
by 5.0% and 13.2% respectively on the MSRA-B dataset and our new dataset
(HKU-IS), and lowering the mean absolute error by 5.7% and 35.1% respectively
on these two datasets.
| [
{
"version": "v1",
"created": "Mon, 30 Mar 2015 13:21:09 GMT"
},
{
"version": "v2",
"created": "Tue, 31 Mar 2015 05:21:02 GMT"
},
{
"version": "v3",
"created": "Fri, 10 Apr 2015 06:40:46 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Li",
"Guanbin",
""
],
[
"Yu",
"Yizhou",
""
]
] | TITLE: Visual Saliency Based on Multiscale Deep Features
ABSTRACT: Visual saliency is a fundamental problem in both cognitive and computational
sciences, including computer vision. In this CVPR 2015 paper, we discover that
a high-quality visual saliency model can be trained with multiscale features
extracted using a popular deep learning architecture, convolutional neural
networks (CNNs), which have had many successes in visual recognition tasks. For
learning such saliency models, we introduce a neural network architecture,
which has fully connected layers on top of CNNs responsible for extracting
features at three different scales. We then propose a refinement method to
enhance the spatial coherence of our saliency results. Finally, aggregating
multiple saliency maps computed for different levels of image segmentation can
further boost the performance, yielding saliency maps better than those
generated from a single segmentation. To promote further research and
evaluation of visual saliency models, we also construct a new large database of
4447 challenging images and their pixelwise saliency annotation. Experimental
results demonstrate that our proposed method is capable of achieving
state-of-the-art performance on all public benchmarks, improving the F-Measure
by 5.0% and 13.2% respectively on the MSRA-B dataset and our new dataset
(HKU-IS), and lowering the mean absolute error by 5.7% and 35.1% respectively
on these two datasets.
| new_dataset | 0.961893 |
1504.02564 | Ragesh Jaiswal | Anup Bhattacharya, Ragesh Jaiswal, Amit Kumar | Faster Algorithms for the Constrained k-means Problem | null | null | null | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The classical center based clustering problems such as
$k$-means/median/center assume that the optimal clusters satisfy the locality
property that the points in the same cluster are close to each other. A number
of clustering problems arise in machine learning where the optimal clusters do
not follow such a locality property. Consider a variant of the $k$-means
problem that may be regarded as a general version of such problems. Here, the
optimal clusters $O_1, ..., O_k$ are an arbitrary partition of the dataset and
the goal is to output $k$-centers $c_1, ..., c_k$ such that the objective
function $\sum_{i=1}^{k} \sum_{x \in O_{i}} ||x - c_{i}||^2$ is minimized. It
is not difficult to argue that any algorithm (without knowing the optimal
clusters) that outputs a single set of $k$ centers, will not behave well as far
as optimizing the above objective function is concerned. However, this does not
rule out the existence of algorithms that output a list of such $k$ centers
such that at least one of these $k$ centers behaves well. Given an error
parameter $\varepsilon > 0$, let $\ell$ denote the size of the smallest list of
$k$-centers such that at least one of the $k$-centers gives a $(1+\varepsilon)$
approximation w.r.t. the objective function above. In this paper, we show an
upper bound on $\ell$ by giving a randomized algorithm that outputs a list of
$2^{\tilde{O}(k/\varepsilon)}$ $k$-centers. We also give a closely matching
lower bound of $2^{\tilde{\Omega}(k/\sqrt{\varepsilon})}$. Moreover, our
algorithm runs in time $O \left(n d \cdot 2^{\tilde{O}(k/\varepsilon)}
\right)$. This is a significant improvement over the previous result of Ding
and Xu who gave an algorithm with running time $O \left(n d \cdot (\log{n})^{k}
\cdot 2^{poly(k/\varepsilon)} \right)$ and output a list of size $O
\left((\log{n})^k \cdot 2^{poly(k/\varepsilon)} \right)$.
| [
{
"version": "v1",
"created": "Fri, 10 Apr 2015 07:03:58 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Bhattacharya",
"Anup",
""
],
[
"Jaiswal",
"Ragesh",
""
],
[
"Kumar",
"Amit",
""
]
] | TITLE: Faster Algorithms for the Constrained k-means Problem
ABSTRACT: The classical center based clustering problems such as
$k$-means/median/center assume that the optimal clusters satisfy the locality
property that the points in the same cluster are close to each other. A number
of clustering problems arise in machine learning where the optimal clusters do
not follow such a locality property. Consider a variant of the $k$-means
problem that may be regarded as a general version of such problems. Here, the
optimal clusters $O_1, ..., O_k$ are an arbitrary partition of the dataset and
the goal is to output $k$-centers $c_1, ..., c_k$ such that the objective
function $\sum_{i=1}^{k} \sum_{x \in O_{i}} ||x - c_{i}||^2$ is minimized. It
is not difficult to argue that any algorithm (without knowing the optimal
clusters) that outputs a single set of $k$ centers, will not behave well as far
as optimizing the above objective function is concerned. However, this does not
rule out the existence of algorithms that output a list of such $k$ centers
such that at least one of these $k$ centers behaves well. Given an error
parameter $\varepsilon > 0$, let $\ell$ denote the size of the smallest list of
$k$-centers such that at least one of the $k$-centers gives a $(1+\varepsilon)$
approximation w.r.t. the objective function above. In this paper, we show an
upper bound on $\ell$ by giving a randomized algorithm that outputs a list of
$2^{\tilde{O}(k/\varepsilon)}$ $k$-centers. We also give a closely matching
lower bound of $2^{\tilde{\Omega}(k/\sqrt{\varepsilon})}$. Moreover, our
algorithm runs in time $O \left(n d \cdot 2^{\tilde{O}(k/\varepsilon)}
\right)$. This is a significant improvement over the previous result of Ding
and Xu who gave an algorithm with running time $O \left(n d \cdot (\log{n})^{k}
\cdot 2^{poly(k/\varepsilon)} \right)$ and output a list of size $O
\left((\log{n})^k \cdot 2^{poly(k/\varepsilon)} \right)$.
| no_new_dataset | 0.946101 |
1504.02696 | Xinyan Yan | Xinyan Yan, Vadim Indelman, Byron Boots | Incremental Sparse GP Regression for Continuous-time Trajectory
Estimation & Mapping | 10 pages, 10 figures | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent work on simultaneous trajectory estimation and mapping (STEAM) for
mobile robots has found success by representing the trajectory as a Gaussian
process. Gaussian processes can represent a continuous-time trajectory,
elegantly handle asynchronous and sparse measurements, and allow the robot to
query the trajectory to recover its estimated position at any time of interest.
A major drawback of this approach is that STEAM is formulated as a batch
estimation problem. In this paper we provide the critical extensions necessary
to transform the existing batch algorithm into an extremely efficient
incremental algorithm. In particular, we are able to vastly speed up the
solution time through efficient variable reordering and incremental sparse
updates, which we believe will greatly increase the practicality of Gaussian
process methods for robot mapping and localization. Finally, we demonstrate the
approach and its advantages on both synthetic and real datasets.
| [
{
"version": "v1",
"created": "Fri, 10 Apr 2015 14:47:25 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Yan",
"Xinyan",
""
],
[
"Indelman",
"Vadim",
""
],
[
"Boots",
"Byron",
""
]
] | TITLE: Incremental Sparse GP Regression for Continuous-time Trajectory
Estimation & Mapping
ABSTRACT: Recent work on simultaneous trajectory estimation and mapping (STEAM) for
mobile robots has found success by representing the trajectory as a Gaussian
process. Gaussian processes can represent a continuous-time trajectory,
elegantly handle asynchronous and sparse measurements, and allow the robot to
query the trajectory to recover its estimated position at any time of interest.
A major drawback of this approach is that STEAM is formulated as a batch
estimation problem. In this paper we provide the critical extensions necessary
to transform the existing batch algorithm into an extremely efficient
incremental algorithm. In particular, we are able to vastly speed up the
solution time through efficient variable reordering and incremental sparse
updates, which we believe will greatly increase the practicality of Gaussian
process methods for robot mapping and localization. Finally, we demonstrate the
approach and its advantages on both synthetic and real datasets.
| no_new_dataset | 0.946646 |
1504.02764 | Roozbeh Mottaghi | Roozbeh Mottaghi, Yu Xiang, Silvio Savarese | A Coarse-to-Fine Model for 3D Pose Estimation and Sub-category
Recognition | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite the fact that object detection, 3D pose estimation, and sub-category
recognition are highly correlated tasks, they are usually addressed
independently from each other because of the huge space of parameters. To
jointly model all of these tasks, we propose a coarse-to-fine hierarchical
representation, where each level of the hierarchy represents objects at a
different level of granularity. The hierarchical representation prevents
performance loss, which is often caused by the increase in the number of
parameters (as we consider more tasks to model), and the joint modelling
enables resolving ambiguities that exist in independent modelling of these
tasks. We augment PASCAL3D+ dataset with annotations for these tasks and show
that our hierarchical model is effective in joint modelling of object
detection, 3D pose estimation, and sub-category recognition.
| [
{
"version": "v1",
"created": "Fri, 10 Apr 2015 19:18:59 GMT"
}
] | 2015-04-13T00:00:00 | [
[
"Mottaghi",
"Roozbeh",
""
],
[
"Xiang",
"Yu",
""
],
[
"Savarese",
"Silvio",
""
]
] | TITLE: A Coarse-to-Fine Model for 3D Pose Estimation and Sub-category
Recognition
ABSTRACT: Despite the fact that object detection, 3D pose estimation, and sub-category
recognition are highly correlated tasks, they are usually addressed
independently from each other because of the huge space of parameters. To
jointly model all of these tasks, we propose a coarse-to-fine hierarchical
representation, where each level of the hierarchy represents objects at a
different level of granularity. The hierarchical representation prevents
performance loss, which is often caused by the increase in the number of
parameters (as we consider more tasks to model), and the joint modelling
enables resolving ambiguities that exist in independent modelling of these
tasks. We augment PASCAL3D+ dataset with annotations for these tasks and show
that our hierarchical model is effective in joint modelling of object
detection, 3D pose estimation, and sub-category recognition.
| no_new_dataset | 0.947088 |
1408.1054 | Wojciech Czarnecki | Wojciech Marian Czarnecki, Jacek Tabor | Multithreshold Entropy Linear Classifier | null | null | 10.1016/j.eswa.2015.03.007 | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Linear classifiers separate the data with a hyperplane. In this paper we
focus on the novel method of construction of multithreshold linear classifier,
which separates the data with multiple parallel hyperplanes. Proposed model is
based on the information theory concepts -- namely Renyi's quadratic entropy
and Cauchy-Schwarz divergence.
We begin with some general properties, including data scale invariance. Then
we prove that our method is a multithreshold large margin classifier, which
shows the analogy to the SVM, while in the same time works with much broader
class of hypotheses. What is also interesting, proposed method is aimed at the
maximization of the balanced quality measure (such as Matthew's Correlation
Coefficient) as opposed to very common maximization of the accuracy. This
feature comes directly from the optimization problem statement and is further
confirmed by the experiments on the UCI datasets.
It appears, that our Multithreshold Entropy Linear Classifier (MELC) obtaines
similar or higher scores than the ones given by SVM on both synthetic and real
data. We show how proposed approach can be benefitial for the cheminformatics
in the task of ligands activity prediction, where despite better classification
results, MELC gives some additional insight into the data structure (classes of
underrepresented chemical compunds).
| [
{
"version": "v1",
"created": "Mon, 4 Aug 2014 18:01:29 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Czarnecki",
"Wojciech Marian",
""
],
[
"Tabor",
"Jacek",
""
]
] | TITLE: Multithreshold Entropy Linear Classifier
ABSTRACT: Linear classifiers separate the data with a hyperplane. In this paper we
focus on the novel method of construction of multithreshold linear classifier,
which separates the data with multiple parallel hyperplanes. Proposed model is
based on the information theory concepts -- namely Renyi's quadratic entropy
and Cauchy-Schwarz divergence.
We begin with some general properties, including data scale invariance. Then
we prove that our method is a multithreshold large margin classifier, which
shows the analogy to the SVM, while in the same time works with much broader
class of hypotheses. What is also interesting, proposed method is aimed at the
maximization of the balanced quality measure (such as Matthew's Correlation
Coefficient) as opposed to very common maximization of the accuracy. This
feature comes directly from the optimization problem statement and is further
confirmed by the experiments on the UCI datasets.
It appears, that our Multithreshold Entropy Linear Classifier (MELC) obtaines
similar or higher scores than the ones given by SVM on both synthetic and real
data. We show how proposed approach can be benefitial for the cheminformatics
in the task of ligands activity prediction, where despite better classification
results, MELC gives some additional insight into the data structure (classes of
underrepresented chemical compunds).
| no_new_dataset | 0.948965 |
1412.7009 | Jan Rudy | Jan Rudy, Graham Taylor | Generative Class-conditional Autoencoders | null | null | null | null | cs.NE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent work by Bengio et al. (2013) proposes a sampling procedure for
denoising autoencoders which involves learning the transition operator of a
Markov chain. The transition operator is typically unimodal, which limits its
capacity to model complex data. In order to perform efficient sampling from
conditional distributions, we extend this work, both theoretically and
algorithmically, to gated autoencoders (Memisevic, 2013), The proposed model is
able to generate convincing class-conditional samples when trained on both the
MNIST and TFD datasets.
| [
{
"version": "v1",
"created": "Mon, 22 Dec 2014 14:57:05 GMT"
},
{
"version": "v2",
"created": "Sat, 28 Feb 2015 00:16:55 GMT"
},
{
"version": "v3",
"created": "Thu, 9 Apr 2015 01:54:33 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Rudy",
"Jan",
""
],
[
"Taylor",
"Graham",
""
]
] | TITLE: Generative Class-conditional Autoencoders
ABSTRACT: Recent work by Bengio et al. (2013) proposes a sampling procedure for
denoising autoencoders which involves learning the transition operator of a
Markov chain. The transition operator is typically unimodal, which limits its
capacity to model complex data. In order to perform efficient sampling from
conditional distributions, we extend this work, both theoretically and
algorithmically, to gated autoencoders (Memisevic, 2013), The proposed model is
able to generate convincing class-conditional samples when trained on both the
MNIST and TFD datasets.
| no_new_dataset | 0.949809 |
1502.01423 | Chen Fang | Chen Fang, Hailin Jin, Jianchao Yang, Zhe Lin | Collaborative Feature Learning from Social Media | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Image feature representation plays an essential role in image recognition and
related tasks. The current state-of-the-art feature learning paradigm is
supervised learning from labeled data. However, this paradigm requires
large-scale category labels, which limits its applicability to domains where
labels are hard to obtain. In this paper, we propose a new data-driven feature
learning paradigm which does not rely on category labels. Instead, we learn
from user behavior data collected on social media. Concretely, we use the image
relationship discovered in the latent space from the user behavior data to
guide the image feature learning. We collect a large-scale image and user
behavior dataset from Behance.net. The dataset consists of 1.9 million images
and over 300 million view records from 1.9 million users. We validate our
feature learning paradigm on this dataset and find that the learned feature
significantly outperforms the state-of-the-art image features in learning
better image similarities. We also show that the learned feature performs
competitively on various recognition benchmarks.
| [
{
"version": "v1",
"created": "Thu, 5 Feb 2015 03:32:19 GMT"
},
{
"version": "v2",
"created": "Sat, 4 Apr 2015 19:27:33 GMT"
},
{
"version": "v3",
"created": "Thu, 9 Apr 2015 18:36:54 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Fang",
"Chen",
""
],
[
"Jin",
"Hailin",
""
],
[
"Yang",
"Jianchao",
""
],
[
"Lin",
"Zhe",
""
]
] | TITLE: Collaborative Feature Learning from Social Media
ABSTRACT: Image feature representation plays an essential role in image recognition and
related tasks. The current state-of-the-art feature learning paradigm is
supervised learning from labeled data. However, this paradigm requires
large-scale category labels, which limits its applicability to domains where
labels are hard to obtain. In this paper, we propose a new data-driven feature
learning paradigm which does not rely on category labels. Instead, we learn
from user behavior data collected on social media. Concretely, we use the image
relationship discovered in the latent space from the user behavior data to
guide the image feature learning. We collect a large-scale image and user
behavior dataset from Behance.net. The dataset consists of 1.9 million images
and over 300 million view records from 1.9 million users. We validate our
feature learning paradigm on this dataset and find that the learned feature
significantly outperforms the state-of-the-art image features in learning
better image similarities. We also show that the learned feature performs
competitively on various recognition benchmarks.
| new_dataset | 0.91181 |
1502.03671 | R\'emi Lebret | R\'emi Lebret, Pedro O. Pinheiro, Ronan Collobert | Phrase-based Image Captioning | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generating a novel textual description of an image is an interesting problem
that connects computer vision and natural language processing. In this paper,
we present a simple model that is able to generate descriptive sentences given
a sample image. This model has a strong focus on the syntax of the
descriptions. We train a purely bilinear model that learns a metric between an
image representation (generated from a previously trained Convolutional Neural
Network) and phrases that are used to described them. The system is then able
to infer phrases from a given image sample. Based on caption syntax statistics,
we propose a simple language model that can produce relevant descriptions for a
given test image using the phrases inferred. Our approach, which is
considerably simpler than state-of-the-art models, achieves comparable results
in two popular datasets for the task: Flickr30k and the recently proposed
Microsoft COCO.
| [
{
"version": "v1",
"created": "Thu, 12 Feb 2015 14:17:15 GMT"
},
{
"version": "v2",
"created": "Thu, 9 Apr 2015 09:48:52 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Lebret",
"Rémi",
""
],
[
"Pinheiro",
"Pedro O.",
""
],
[
"Collobert",
"Ronan",
""
]
] | TITLE: Phrase-based Image Captioning
ABSTRACT: Generating a novel textual description of an image is an interesting problem
that connects computer vision and natural language processing. In this paper,
we present a simple model that is able to generate descriptive sentences given
a sample image. This model has a strong focus on the syntax of the
descriptions. We train a purely bilinear model that learns a metric between an
image representation (generated from a previously trained Convolutional Neural
Network) and phrases that are used to described them. The system is then able
to infer phrases from a given image sample. Based on caption syntax statistics,
we propose a simple language model that can produce relevant descriptions for a
given test image using the phrases inferred. Our approach, which is
considerably simpler than state-of-the-art models, achieves comparable results
in two popular datasets for the task: Flickr30k and the recently proposed
Microsoft COCO.
| no_new_dataset | 0.949201 |
1504.02147 | Thomas Goldstein | Tom Goldstein, Gavin Taylor, Kawika Barabin, Kent Sayre | Unwrapping ADMM: Efficient Distributed Computing via Transpose Reduction | null | null | null | null | cs.DC cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent approaches to distributed model fitting rely heavily on consensus
ADMM, where each node solves small sub-problems using only local data. We
propose iterative methods that solve {\em global} sub-problems over an entire
distributed dataset. This is possible using transpose reduction strategies that
allow a single node to solve least-squares over massive datasets without
putting all the data in one place. This results in simple iterative methods
that avoid the expensive inner loops required for consensus methods. To
demonstrate the efficiency of this approach, we fit linear classifiers and
sparse linear models to datasets over 5 Tb in size using a distributed
implementation with over 7000 cores in far less time than previous approaches.
| [
{
"version": "v1",
"created": "Wed, 8 Apr 2015 22:35:18 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Goldstein",
"Tom",
""
],
[
"Taylor",
"Gavin",
""
],
[
"Barabin",
"Kawika",
""
],
[
"Sayre",
"Kent",
""
]
] | TITLE: Unwrapping ADMM: Efficient Distributed Computing via Transpose Reduction
ABSTRACT: Recent approaches to distributed model fitting rely heavily on consensus
ADMM, where each node solves small sub-problems using only local data. We
propose iterative methods that solve {\em global} sub-problems over an entire
distributed dataset. This is possible using transpose reduction strategies that
allow a single node to solve least-squares over massive datasets without
putting all the data in one place. This results in simple iterative methods
that avoid the expensive inner loops required for consensus methods. To
demonstrate the efficiency of this approach, we fit linear classifiers and
sparse linear models to datasets over 5 Tb in size using a distributed
implementation with over 7000 cores in far less time than previous approaches.
| no_new_dataset | 0.946498 |
1504.02305 | Haewoon Kwak | Haewoon Kwak and Jeremy Blackburn and Seungyeop Han | Exploring Cyberbullying and Other Toxic Behavior in Team Competition
Online Games | CHI'15 | null | null | null | cs.CY cs.HC cs.MM cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we explore cyberbullying and other toxic behavior in team
competition online games. Using a dataset of over 10 million player reports on
1.46 million toxic players along with corresponding crowdsourced decisions, we
test several hypotheses drawn from theories explaining toxic behavior. Besides
providing large-scale, empirical based understanding of toxic behavior, our
work can be used as a basis for building systems to detect, prevent, and
counter-act toxic behavior.
| [
{
"version": "v1",
"created": "Thu, 9 Apr 2015 13:33:58 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Kwak",
"Haewoon",
""
],
[
"Blackburn",
"Jeremy",
""
],
[
"Han",
"Seungyeop",
""
]
] | TITLE: Exploring Cyberbullying and Other Toxic Behavior in Team Competition
Online Games
ABSTRACT: In this work we explore cyberbullying and other toxic behavior in team
competition online games. Using a dataset of over 10 million player reports on
1.46 million toxic players along with corresponding crowdsourced decisions, we
test several hypotheses drawn from theories explaining toxic behavior. Besides
providing large-scale, empirical based understanding of toxic behavior, our
work can be used as a basis for building systems to detect, prevent, and
counter-act toxic behavior.
| no_new_dataset | 0.835819 |
1504.02340 | Wongun Choi | Wongun Choi | Near-Online Multi-target Tracking with Aggregated Local Flow Descriptor | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we focus on the two key aspects of multiple target tracking
problem: 1) designing an accurate affinity measure to associate detections and
2) implementing an efficient and accurate (near) online multiple target
tracking algorithm. As the first contribution, we introduce a novel Aggregated
Local Flow Descriptor (ALFD) that encodes the relative motion pattern between a
pair of temporally distant detections using long term interest point
trajectories (IPTs). Leveraging on the IPTs, the ALFD provides a robust
affinity measure for estimating the likelihood of matching detections
regardless of the application scenarios. As another contribution, we present a
Near-Online Multi-target Tracking (NOMT) algorithm. The tracking problem is
formulated as a data-association between targets and detections in a temporal
window, that is performed repeatedly at every frame. While being efficient,
NOMT achieves robustness via integrating multiple cues including ALFD metric,
target dynamics, appearance similarity, and long term trajectory regularization
into the model. Our ablative analysis verifies the superiority of the ALFD
metric over the other conventional affinity metrics. We run a comprehensive
experimental evaluation on two challenging tracking datasets, KITTI and MOT
datasets. The NOMT method combined with ALFD metric achieves the best accuracy
in both datasets with significant margins (about 10% higher MOTA) over the
state-of-the-arts.
| [
{
"version": "v1",
"created": "Thu, 9 Apr 2015 14:57:32 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Choi",
"Wongun",
""
]
] | TITLE: Near-Online Multi-target Tracking with Aggregated Local Flow Descriptor
ABSTRACT: In this paper, we focus on the two key aspects of multiple target tracking
problem: 1) designing an accurate affinity measure to associate detections and
2) implementing an efficient and accurate (near) online multiple target
tracking algorithm. As the first contribution, we introduce a novel Aggregated
Local Flow Descriptor (ALFD) that encodes the relative motion pattern between a
pair of temporally distant detections using long term interest point
trajectories (IPTs). Leveraging on the IPTs, the ALFD provides a robust
affinity measure for estimating the likelihood of matching detections
regardless of the application scenarios. As another contribution, we present a
Near-Online Multi-target Tracking (NOMT) algorithm. The tracking problem is
formulated as a data-association between targets and detections in a temporal
window, that is performed repeatedly at every frame. While being efficient,
NOMT achieves robustness via integrating multiple cues including ALFD metric,
target dynamics, appearance similarity, and long term trajectory regularization
into the model. Our ablative analysis verifies the superiority of the ALFD
metric over the other conventional affinity metrics. We run a comprehensive
experimental evaluation on two challenging tracking datasets, KITTI and MOT
datasets. The NOMT method combined with ALFD metric achieves the best accuracy
in both datasets with significant margins (about 10% higher MOTA) over the
state-of-the-arts.
| no_new_dataset | 0.945399 |
1504.02356 | Xavier Gir\'o-i-Nieto | Eva Mohedano, Amaia Salvador, Sergi Porta, Xavier Gir\'o-i-Nieto,
Graham Healy, Kevin McGuinness, Noel O'Connor and Alan F. Smeaton | Exploring EEG for Object Detection and Retrieval | This preprint is the full version of a short paper accepted in the
ACM International Conference on Multimedia Retrieval (ICMR) 2015 (Shanghai,
China) | null | null | null | cs.HC cs.CV cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper explores the potential for using Brain Computer Interfaces (BCI)
as a relevance feedback mechanism in content-based image retrieval. We
investigate if it is possible to capture useful EEG signals to detect if
relevant objects are present in a dataset of realistic and complex images. We
perform several experiments using a rapid serial visual presentation (RSVP) of
images at different rates (5Hz and 10Hz) on 8 users with different degrees of
familiarization with BCI and the dataset. We then use the feedback from the BCI
and mouse-based interfaces to retrieve localized objects in a subset of TRECVid
images. We show that it is indeed possible to detect such objects in complex
images and, also, that users with previous knowledge on the dataset or
experience with the RSVP outperform others. When the users have limited time to
annotate the images (100 seconds in our experiments) both interfaces are
comparable in performance. Comparing our best users in a retrieval task, we
found that EEG-based relevance feedback outperforms mouse-based feedback. The
realistic and complex image dataset differentiates our work from previous
studies on EEG for image retrieval.
| [
{
"version": "v1",
"created": "Thu, 9 Apr 2015 15:43:52 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Mohedano",
"Eva",
""
],
[
"Salvador",
"Amaia",
""
],
[
"Porta",
"Sergi",
""
],
[
"Giró-i-Nieto",
"Xavier",
""
],
[
"Healy",
"Graham",
""
],
[
"McGuinness",
"Kevin",
""
],
[
"O'Connor",
"Noel",
""
],
[
"Smeaton",
"Alan F.",
""
]
] | TITLE: Exploring EEG for Object Detection and Retrieval
ABSTRACT: This paper explores the potential for using Brain Computer Interfaces (BCI)
as a relevance feedback mechanism in content-based image retrieval. We
investigate if it is possible to capture useful EEG signals to detect if
relevant objects are present in a dataset of realistic and complex images. We
perform several experiments using a rapid serial visual presentation (RSVP) of
images at different rates (5Hz and 10Hz) on 8 users with different degrees of
familiarization with BCI and the dataset. We then use the feedback from the BCI
and mouse-based interfaces to retrieve localized objects in a subset of TRECVid
images. We show that it is indeed possible to detect such objects in complex
images and, also, that users with previous knowledge on the dataset or
experience with the RSVP outperform others. When the users have limited time to
annotate the images (100 seconds in our experiments) both interfaces are
comparable in performance. Comparing our best users in a retrieval task, we
found that EEG-based relevance feedback outperforms mouse-based feedback. The
realistic and complex image dataset differentiates our work from previous
studies on EEG for image retrieval.
| no_new_dataset | 0.940353 |
1504.02363 | Ke Zhang | Ke Zhang, Konstantinos Pelechrinis, Theodoros Lappas | Analyzing and Modeling Special Offer Campaigns in Location-based Social
Networks | in The 9th International AAAI Conference on Web and Social Media
(ICWSM 2015) | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The proliferation of mobile handheld devices in combination with the
technological advancements in mobile computing has led to a number of
innovative services that make use of the location information available on such
devices. Traditional yellow pages websites have now moved to mobile platforms,
giving the opportunity to local businesses and potential, near-by, customers to
connect. These platforms can offer an affordable advertisement channel to local
businesses. One of the mechanisms offered by location-based social networks
(LBSNs) allows businesses to provide special offers to their customers that
connect through the platform. We collect a large time-series dataset from
approximately 14 million venues on Foursquare and analyze the performance of
such campaigns using randomization techniques and (non-parametric) hypothesis
testing with statistical bootstrapping. Our main finding indicates that this
type of promotions are not as effective as anecdote success stories might
suggest. Finally, we design classifiers by extracting three different types of
features that are able to provide an educated decision on whether a special
offer campaign for a local business will succeed or not both in short and long
term.
| [
{
"version": "v1",
"created": "Thu, 9 Apr 2015 16:26:40 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Zhang",
"Ke",
""
],
[
"Pelechrinis",
"Konstantinos",
""
],
[
"Lappas",
"Theodoros",
""
]
] | TITLE: Analyzing and Modeling Special Offer Campaigns in Location-based Social
Networks
ABSTRACT: The proliferation of mobile handheld devices in combination with the
technological advancements in mobile computing has led to a number of
innovative services that make use of the location information available on such
devices. Traditional yellow pages websites have now moved to mobile platforms,
giving the opportunity to local businesses and potential, near-by, customers to
connect. These platforms can offer an affordable advertisement channel to local
businesses. One of the mechanisms offered by location-based social networks
(LBSNs) allows businesses to provide special offers to their customers that
connect through the platform. We collect a large time-series dataset from
approximately 14 million venues on Foursquare and analyze the performance of
such campaigns using randomization techniques and (non-parametric) hypothesis
testing with statistical bootstrapping. Our main finding indicates that this
type of promotions are not as effective as anecdote success stories might
suggest. Finally, we design classifiers by extracting three different types of
features that are able to provide an educated decision on whether a special
offer campaign for a local business will succeed or not both in short and long
term.
| no_new_dataset | 0.933613 |
1504.02380 | Spencer Wheatley Mr. | Spencer Wheatley, Benjamin Sovacool, and Didier Sornette | Of Disasters and Dragon Kings: A Statistical Analysis of Nuclear Power
Incidents & Accidents | null | null | null | null | physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We provide, and perform a risk theoretic statistical analysis of, a dataset
that is 75 percent larger than the previous best dataset on nuclear incidents
and accidents, comparing three measures of severity: INES (International
Nuclear Event Scale), radiation released, and damage dollar losses. The annual
rate of nuclear accidents, with size above 20 Million US$, per plant, decreased
from the 1950s until dropping significantly after Chernobyl (April, 1986). The
rate is now roughly stable at 0.002 to 0.003, i.e., around 1 event per year
across the current fleet. The distribution of damage values changed after Three
Mile Island (TMI; March, 1979), where moderate damages were suppressed but the
tail became very heavy, being described by a Pareto distribution with tail
index 0.55. Further, there is a runaway disaster regime, associated with the
"dragon-king" phenomenon, amplifying the risk of extreme damage. In fact, the
damage of the largest event (Fukushima; March, 2011) is equal to 60 percent of
the total damage of all 174 accidents in our database since 1946. In dollar
losses we compute a 50% chance that (i) a Fukushima event (or larger) occurs in
the next 50 years, (ii) a Chernobyl event (or larger) occurs in the next 27
years and (iii) a TMI event (or larger) occurs in the next 10 years. Finally,
we find that the INES scale is inconsistent. To be consistent with damage, the
Fukushima disaster would need to have an INES level of 11, rather than the
maximum of 7.
| [
{
"version": "v1",
"created": "Tue, 7 Apr 2015 18:50:33 GMT"
}
] | 2015-04-10T00:00:00 | [
[
"Wheatley",
"Spencer",
""
],
[
"Sovacool",
"Benjamin",
""
],
[
"Sornette",
"Didier",
""
]
] | TITLE: Of Disasters and Dragon Kings: A Statistical Analysis of Nuclear Power
Incidents & Accidents
ABSTRACT: We provide, and perform a risk theoretic statistical analysis of, a dataset
that is 75 percent larger than the previous best dataset on nuclear incidents
and accidents, comparing three measures of severity: INES (International
Nuclear Event Scale), radiation released, and damage dollar losses. The annual
rate of nuclear accidents, with size above 20 Million US$, per plant, decreased
from the 1950s until dropping significantly after Chernobyl (April, 1986). The
rate is now roughly stable at 0.002 to 0.003, i.e., around 1 event per year
across the current fleet. The distribution of damage values changed after Three
Mile Island (TMI; March, 1979), where moderate damages were suppressed but the
tail became very heavy, being described by a Pareto distribution with tail
index 0.55. Further, there is a runaway disaster regime, associated with the
"dragon-king" phenomenon, amplifying the risk of extreme damage. In fact, the
damage of the largest event (Fukushima; March, 2011) is equal to 60 percent of
the total damage of all 174 accidents in our database since 1946. In dollar
losses we compute a 50% chance that (i) a Fukushima event (or larger) occurs in
the next 50 years, (ii) a Chernobyl event (or larger) occurs in the next 27
years and (iii) a TMI event (or larger) occurs in the next 10 years. Finally,
we find that the INES scale is inconsistent. To be consistent with damage, the
Fukushima disaster would need to have an INES level of 11, rather than the
maximum of 7.
| no_new_dataset | 0.929376 |
1504.01781 | Tiep Mai | Tiep Mai, Deepak Ajwani and Alessandra Sala | Profiling user activities with minimal traffic traces | null | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Understanding user behavior is essential to personalize and enrich a user's
online experience. While there are significant benefits to be accrued from the
pursuit of personalized services based on a fine-grained behavioral analysis,
care must be taken to address user privacy concerns. In this paper, we consider
the use of web traces with truncated URLs - each URL is trimmed to only contain
the web domain - for this purpose. While such truncation removes the
fine-grained sensitive information, it also strips the data of many features
that are crucial to the profiling of user activity. We show how to overcome the
severe handicap of lack of crucial features for the purpose of filtering out
the URLs representing a user activity from the noisy network traffic trace
(including advertisement, spam, analytics, webscripts) with high accuracy. This
activity profiling with truncated URLs enables the network operators to provide
personalized services while mitigating privacy concerns by storing and sharing
only truncated traffic traces.
In order to offset the accuracy loss due to truncation, our statistical
methodology leverages specialized features extracted from a group of
consecutive URLs that represent a micro user action like web click, chat reply,
etc., which we call bursts. These bursts, in turn, are detected by a novel
algorithm which is based on our observed characteristics of the inter-arrival
time of HTTP records. We present an extensive experimental evaluation on a real
dataset of mobile web traces, consisting of more than 130 million records,
representing the browsing activities of 10,000 users over a period of 30 days.
Our results show that the proposed methodology achieves around 90% accuracy in
segregating URLs representing user activities from non-representative URLs.
| [
{
"version": "v1",
"created": "Tue, 7 Apr 2015 23:29:23 GMT"
}
] | 2015-04-09T00:00:00 | [
[
"Mai",
"Tiep",
""
],
[
"Ajwani",
"Deepak",
""
],
[
"Sala",
"Alessandra",
""
]
] | TITLE: Profiling user activities with minimal traffic traces
ABSTRACT: Understanding user behavior is essential to personalize and enrich a user's
online experience. While there are significant benefits to be accrued from the
pursuit of personalized services based on a fine-grained behavioral analysis,
care must be taken to address user privacy concerns. In this paper, we consider
the use of web traces with truncated URLs - each URL is trimmed to only contain
the web domain - for this purpose. While such truncation removes the
fine-grained sensitive information, it also strips the data of many features
that are crucial to the profiling of user activity. We show how to overcome the
severe handicap of lack of crucial features for the purpose of filtering out
the URLs representing a user activity from the noisy network traffic trace
(including advertisement, spam, analytics, webscripts) with high accuracy. This
activity profiling with truncated URLs enables the network operators to provide
personalized services while mitigating privacy concerns by storing and sharing
only truncated traffic traces.
In order to offset the accuracy loss due to truncation, our statistical
methodology leverages specialized features extracted from a group of
consecutive URLs that represent a micro user action like web click, chat reply,
etc., which we call bursts. These bursts, in turn, are detected by a novel
algorithm which is based on our observed characteristics of the inter-arrival
time of HTTP records. We present an extensive experimental evaluation on a real
dataset of mobile web traces, consisting of more than 130 million records,
representing the browsing activities of 10,000 users over a period of 30 days.
Our results show that the proposed methodology achieves around 90% accuracy in
segregating URLs representing user activities from non-representative URLs.
| new_dataset | 0.958148 |
1504.01807 | Junbin Gao Professor | Boyue Wang and Yongli Hu and Junbin Gao and Yanfeng Sun and Baocai Yin | Low Rank Representation on Grassmann Manifolds: An Extrinsic Perspective | 9 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many computer vision algorithms employ subspace models to represent data. The
Low-rank representation (LRR) has been successfully applied in subspace
clustering for which data are clustered according to their subspace structures.
The possibility of extending LRR on Grassmann manifold is explored in this
paper. Rather than directly embedding Grassmann manifold into a symmetric
matrix space, an extrinsic view is taken by building the self-representation of
LRR over the tangent space of each Grassmannian point. A new algorithm for
solving the proposed Grassmannian LRR model is designed and implemented.
Several clustering experiments are conducted on handwritten digits dataset,
dynamic texture video clips and YouTube celebrity face video data. The
experimental results show our method outperforms a number of existing methods.
| [
{
"version": "v1",
"created": "Wed, 8 Apr 2015 02:38:04 GMT"
}
] | 2015-04-09T00:00:00 | [
[
"Wang",
"Boyue",
""
],
[
"Hu",
"Yongli",
""
],
[
"Gao",
"Junbin",
""
],
[
"Sun",
"Yanfeng",
""
],
[
"Yin",
"Baocai",
""
]
] | TITLE: Low Rank Representation on Grassmann Manifolds: An Extrinsic Perspective
ABSTRACT: Many computer vision algorithms employ subspace models to represent data. The
Low-rank representation (LRR) has been successfully applied in subspace
clustering for which data are clustered according to their subspace structures.
The possibility of extending LRR on Grassmann manifold is explored in this
paper. Rather than directly embedding Grassmann manifold into a symmetric
matrix space, an extrinsic view is taken by building the self-representation of
LRR over the tangent space of each Grassmannian point. A new algorithm for
solving the proposed Grassmannian LRR model is designed and implemented.
Several clustering experiments are conducted on handwritten digits dataset,
dynamic texture video clips and YouTube celebrity face video data. The
experimental results show our method outperforms a number of existing methods.
| no_new_dataset | 0.947478 |
1504.01840 | Yegor Tkachenko | Yegor Tkachenko | Autonomous CRM Control via CLV Approximation with Deep Reinforcement
Learning in Discrete and Continuous Action Space | 13 pages, 8 figures | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The paper outlines a framework for autonomous control of a CRM (customer
relationship management) system. First, it explores how a modified version of
the widely accepted Recency-Frequency-Monetary Value system of metrics can be
used to define the state space of clients or donors. Second, it describes a
procedure to determine the optimal direct marketing action in discrete and
continuous action space for the given individual, based on his position in the
state space. The procedure involves the use of model-free Q-learning to train a
deep neural network that relates a client's position in the state space to
rewards associated with possible marketing actions. The estimated value
function over the client state space can be interpreted as customer lifetime
value, and thus allows for a quick plug-in estimation of CLV for a given
client. Experimental results are presented, based on KDD Cup 1998 mailing
dataset of donation solicitations.
| [
{
"version": "v1",
"created": "Wed, 8 Apr 2015 06:22:44 GMT"
}
] | 2015-04-09T00:00:00 | [
[
"Tkachenko",
"Yegor",
""
]
] | TITLE: Autonomous CRM Control via CLV Approximation with Deep Reinforcement
Learning in Discrete and Continuous Action Space
ABSTRACT: The paper outlines a framework for autonomous control of a CRM (customer
relationship management) system. First, it explores how a modified version of
the widely accepted Recency-Frequency-Monetary Value system of metrics can be
used to define the state space of clients or donors. Second, it describes a
procedure to determine the optimal direct marketing action in discrete and
continuous action space for the given individual, based on his position in the
state space. The procedure involves the use of model-free Q-learning to train a
deep neural network that relates a client's position in the state space to
rewards associated with possible marketing actions. The estimated value
function over the client state space can be interpreted as customer lifetime
value, and thus allows for a quick plug-in estimation of CLV for a given
client. Experimental results are presented, based on KDD Cup 1998 mailing
dataset of donation solicitations.
| no_new_dataset | 0.947575 |
1504.01942 | Laura Leal Taix\'e | Laura Leal-Taix\'e and Anton Milan and Ian Reid and Stefan Roth and
Konrad Schindler | MOTChallenge 2015: Towards a Benchmark for Multi-Target Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the recent past, the computer vision community has developed centralized
benchmarks for the performance evaluation of a variety of tasks, including
generic object and pedestrian detection, 3D reconstruction, optical flow,
single-object short-term tracking, and stereo estimation. Despite potential
pitfalls of such benchmarks, they have proved to be extremely helpful to
advance the state of the art in the respective area. Interestingly, there has
been rather limited work on the standardization of quantitative benchmarks for
multiple target tracking. One of the few exceptions is the well-known PETS
dataset, targeted primarily at surveillance applications. Despite being widely
used, it is often applied inconsistently, for example involving using different
subsets of the available data, different ways of training the models, or
differing evaluation scripts. This paper describes our work toward a novel
multiple object tracking benchmark aimed to address such issues. We discuss the
challenges of creating such a framework, collecting existing and new data,
gathering state-of-the-art methods to be tested on the datasets, and finally
creating a unified evaluation system. With MOTChallenge we aim to pave the way
toward a unified evaluation framework for a more meaningful quantification of
multi-target tracking.
| [
{
"version": "v1",
"created": "Wed, 8 Apr 2015 12:56:38 GMT"
}
] | 2015-04-09T00:00:00 | [
[
"Leal-Taixé",
"Laura",
""
],
[
"Milan",
"Anton",
""
],
[
"Reid",
"Ian",
""
],
[
"Roth",
"Stefan",
""
],
[
"Schindler",
"Konrad",
""
]
] | TITLE: MOTChallenge 2015: Towards a Benchmark for Multi-Target Tracking
ABSTRACT: In the recent past, the computer vision community has developed centralized
benchmarks for the performance evaluation of a variety of tasks, including
generic object and pedestrian detection, 3D reconstruction, optical flow,
single-object short-term tracking, and stereo estimation. Despite potential
pitfalls of such benchmarks, they have proved to be extremely helpful to
advance the state of the art in the respective area. Interestingly, there has
been rather limited work on the standardization of quantitative benchmarks for
multiple target tracking. One of the few exceptions is the well-known PETS
dataset, targeted primarily at surveillance applications. Despite being widely
used, it is often applied inconsistently, for example involving using different
subsets of the available data, different ways of training the models, or
differing evaluation scripts. This paper describes our work toward a novel
multiple object tracking benchmark aimed to address such issues. We discuss the
challenges of creating such a framework, collecting existing and new data,
gathering state-of-the-art methods to be tested on the datasets, and finally
creating a unified evaluation system. With MOTChallenge we aim to pave the way
toward a unified evaluation framework for a more meaningful quantification of
multi-target tracking.
| no_new_dataset | 0.882276 |
1504.01957 | Tony Tsang | Tony Tsang | Video Contents Prior Storing Server for Optical Access Network | 10 pages. in March 2015, Volume 7. Number 1 International Journal of
Computer Networks & Communications (IJCNC) March 2015, Volume 7 | null | null | null | cs.NI cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | One of the most important multimedia applications is Internet protocol TV
(IPTV) for next-generation networks. IPTV provides triple-play services that
require high-speed access networks with the functions of multicasting and
quality of service (QoS) guarantees. Among optical access networks, Ethernet
passive optical networks (EPONs) are regarded as among the best solutions to
meet higher bandwidth demands. In this paper, we propose a new architecture for
multicasting live IPTV traffic in optical access network. The proposed
mechanism involves assigning a unique logical link identifier to each IPTV
channel. To manage multicasting, a prior storing server in the optical line
terminal (OLT) and in each optical network unit (ONU) is constructed. In this
work, we propose a partial prior storing strategy that considers the changes in
the popularity of the video content segments over time and the access patterns
of the users to compute the utility of the objects in the prior storage. We
also propose to partition the prior storage to avoid the eviction of the
popular objects (those not accessed frequently) by the unpopular ones which are
accessed with higher frequency. The popularity distribution and ageing of
popularity are measured from two online datasets and use the parameters in
simulations. Simulation results show that our proposed architecture can improve
the system performance and QoS parameters in terms of packet delay, jitter and
packet loss.
| [
{
"version": "v1",
"created": "Wed, 8 Apr 2015 13:29:12 GMT"
}
] | 2015-04-09T00:00:00 | [
[
"Tsang",
"Tony",
""
]
] | TITLE: Video Contents Prior Storing Server for Optical Access Network
ABSTRACT: One of the most important multimedia applications is Internet protocol TV
(IPTV) for next-generation networks. IPTV provides triple-play services that
require high-speed access networks with the functions of multicasting and
quality of service (QoS) guarantees. Among optical access networks, Ethernet
passive optical networks (EPONs) are regarded as among the best solutions to
meet higher bandwidth demands. In this paper, we propose a new architecture for
multicasting live IPTV traffic in optical access network. The proposed
mechanism involves assigning a unique logical link identifier to each IPTV
channel. To manage multicasting, a prior storing server in the optical line
terminal (OLT) and in each optical network unit (ONU) is constructed. In this
work, we propose a partial prior storing strategy that considers the changes in
the popularity of the video content segments over time and the access patterns
of the users to compute the utility of the objects in the prior storage. We
also propose to partition the prior storage to avoid the eviction of the
popular objects (those not accessed frequently) by the unpopular ones which are
accessed with higher frequency. The popularity distribution and ageing of
popularity are measured from two online datasets and use the parameters in
simulations. Simulation results show that our proposed architecture can improve
the system performance and QoS parameters in terms of packet delay, jitter and
packet loss.
| no_new_dataset | 0.953449 |
1411.3715 | Daniele Barchiesi | Daniele Barchiesi, Dimitrios Giannoulis, Dan Stowell, Mark D. Plumbley | Acoustic Scene Classification | null | IEEE Signal Processing Magazine 32(3) (May 2015) 16-34 | 10.1109/MSP.2014.2326181 | null | cs.SD cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this article we present an account of the state-of-the-art in acoustic
scene classification (ASC), the task of classifying environments from the
sounds they produce. Starting from a historical review of previous research in
this area, we define a general framework for ASC and present different imple-
mentations of its components. We then describe a range of different algorithms
submitted for a data challenge that was held to provide a general and fair
benchmark for ASC techniques. The dataset recorded for this purpose is
presented, along with the performance metrics that are used to evaluate the
algorithms and statistical significance tests to compare the submitted methods.
We use a baseline method that employs MFCCS, GMMS and a maximum likelihood
criterion as a benchmark, and only find sufficient evidence to conclude that
three algorithms significantly outperform it. We also evaluate the human
classification accuracy in performing a similar classification task. The best
performing algorithm achieves a mean accuracy that matches the median accuracy
obtained by humans, and common pairs of classes are misclassified by both
computers and humans. However, all acoustic scenes are correctly classified by
at least some individuals, while there are scenes that are misclassified by all
algorithms.
| [
{
"version": "v1",
"created": "Thu, 13 Nov 2014 16:03:09 GMT"
}
] | 2015-04-08T00:00:00 | [
[
"Barchiesi",
"Daniele",
""
],
[
"Giannoulis",
"Dimitrios",
""
],
[
"Stowell",
"Dan",
""
],
[
"Plumbley",
"Mark D.",
""
]
] | TITLE: Acoustic Scene Classification
ABSTRACT: In this article we present an account of the state-of-the-art in acoustic
scene classification (ASC), the task of classifying environments from the
sounds they produce. Starting from a historical review of previous research in
this area, we define a general framework for ASC and present different imple-
mentations of its components. We then describe a range of different algorithms
submitted for a data challenge that was held to provide a general and fair
benchmark for ASC techniques. The dataset recorded for this purpose is
presented, along with the performance metrics that are used to evaluate the
algorithms and statistical significance tests to compare the submitted methods.
We use a baseline method that employs MFCCS, GMMS and a maximum likelihood
criterion as a benchmark, and only find sufficient evidence to conclude that
three algorithms significantly outperform it. We also evaluate the human
classification accuracy in performing a similar classification task. The best
performing algorithm achieves a mean accuracy that matches the median accuracy
obtained by humans, and common pairs of classes are misclassified by both
computers and humans. However, all acoustic scenes are correctly classified by
at least some individuals, while there are scenes that are misclassified by all
algorithms.
| new_dataset | 0.973062 |
1412.6615 | Levent Sagun | Levent Sagun, V. Ugur Guney, Gerard Ben Arous, Yann LeCun | Explorations on high dimensional landscapes | 11 pages, 8 figures, workshop contribution at ICLR 2015 | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Finding minima of a real valued non-convex function over a high dimensional
space is a major challenge in science. We provide evidence that some such
functions that are defined on high dimensional domains have a narrow band of
values whose pre-image contains the bulk of its critical points. This is in
contrast with the low dimensional picture in which this band is wide. Our
simulations agree with the previous theoretical work on spin glasses that
proves the existence of such a band when the dimension of the domain tends to
infinity. Furthermore our experiments on teacher-student networks with the
MNIST dataset establish a similar phenomenon in deep networks. We finally
observe that both the gradient descent and the stochastic gradient descent
methods can reach this level within the same number of steps.
| [
{
"version": "v1",
"created": "Sat, 20 Dec 2014 06:57:12 GMT"
},
{
"version": "v2",
"created": "Thu, 25 Dec 2014 01:29:56 GMT"
},
{
"version": "v3",
"created": "Mon, 2 Mar 2015 10:08:16 GMT"
},
{
"version": "v4",
"created": "Mon, 6 Apr 2015 21:47:50 GMT"
}
] | 2015-04-08T00:00:00 | [
[
"Sagun",
"Levent",
""
],
[
"Guney",
"V. Ugur",
""
],
[
"Arous",
"Gerard Ben",
""
],
[
"LeCun",
"Yann",
""
]
] | TITLE: Explorations on high dimensional landscapes
ABSTRACT: Finding minima of a real valued non-convex function over a high dimensional
space is a major challenge in science. We provide evidence that some such
functions that are defined on high dimensional domains have a narrow band of
values whose pre-image contains the bulk of its critical points. This is in
contrast with the low dimensional picture in which this band is wide. Our
simulations agree with the previous theoretical work on spin glasses that
proves the existence of such a band when the dimension of the domain tends to
infinity. Furthermore our experiments on teacher-student networks with the
MNIST dataset establish a similar phenomenon in deep networks. We finally
observe that both the gradient descent and the stochastic gradient descent
methods can reach this level within the same number of steps.
| no_new_dataset | 0.951414 |
1504.01383 | Cristian Danescu-Niculescu-Mizil | Vlad Niculae, Caroline Suen, Justine Zhang, Cristian
Danescu-Niculescu-Mizil, Jure Leskovec | QUOTUS: The Structure of Political Media Coverage as Revealed by Quoting
Patterns | To appear in the Proceedings of WWW 2015. 11pp, 10 fig. Interactive
visualization, data, and other info available at
http://snap.stanford.edu/quotus/ | null | null | null | cs.CL cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given the extremely large pool of events and stories available, media outlets
need to focus on a subset of issues and aspects to convey to their audience.
Outlets are often accused of exhibiting a systematic bias in this selection
process, with different outlets portraying different versions of reality.
However, in the absence of objective measures and empirical evidence, the
direction and extent of systematicity remains widely disputed.
In this paper we propose a framework based on quoting patterns for
quantifying and characterizing the degree to which media outlets exhibit
systematic bias. We apply this framework to a massive dataset of news articles
spanning the six years of Obama's presidency and all of his speeches, and
reveal that a systematic pattern does indeed emerge from the outlet's quoting
behavior. Moreover, we show that this pattern can be successfully exploited in
an unsupervised prediction setting, to determine which new quotes an outlet
will select to broadcast. By encoding bias patterns in a low-rank space we
provide an analysis of the structure of political media coverage. This reveals
a latent media bias space that aligns surprisingly well with political ideology
and outlet type. A linguistic analysis exposes striking differences across
these latent dimensions, showing how the different types of media outlets
portray different realities even when reporting on the same events. For
example, outlets mapped to the mainstream conservative side of the latent space
focus on quotes that portray a presidential persona disproportionately
characterized by negativity.
| [
{
"version": "v1",
"created": "Mon, 6 Apr 2015 20:00:28 GMT"
}
] | 2015-04-08T00:00:00 | [
[
"Niculae",
"Vlad",
""
],
[
"Suen",
"Caroline",
""
],
[
"Zhang",
"Justine",
""
],
[
"Danescu-Niculescu-Mizil",
"Cristian",
""
],
[
"Leskovec",
"Jure",
""
]
] | TITLE: QUOTUS: The Structure of Political Media Coverage as Revealed by Quoting
Patterns
ABSTRACT: Given the extremely large pool of events and stories available, media outlets
need to focus on a subset of issues and aspects to convey to their audience.
Outlets are often accused of exhibiting a systematic bias in this selection
process, with different outlets portraying different versions of reality.
However, in the absence of objective measures and empirical evidence, the
direction and extent of systematicity remains widely disputed.
In this paper we propose a framework based on quoting patterns for
quantifying and characterizing the degree to which media outlets exhibit
systematic bias. We apply this framework to a massive dataset of news articles
spanning the six years of Obama's presidency and all of his speeches, and
reveal that a systematic pattern does indeed emerge from the outlet's quoting
behavior. Moreover, we show that this pattern can be successfully exploited in
an unsupervised prediction setting, to determine which new quotes an outlet
will select to broadcast. By encoding bias patterns in a low-rank space we
provide an analysis of the structure of political media coverage. This reveals
a latent media bias space that aligns surprisingly well with political ideology
and outlet type. A linguistic analysis exposes striking differences across
these latent dimensions, showing how the different types of media outlets
portray different realities even when reporting on the same events. For
example, outlets mapped to the mainstream conservative side of the latent space
focus on quotes that portray a presidential persona disproportionately
characterized by negativity.
| no_new_dataset | 0.945551 |
1504.01684 | Miao Fan | Miao Fan, Qiang Zhou, Thomas Fang Zheng and Ralph Grishman | Large Margin Nearest Neighbor Embedding for Knowledge Representation | arXiv admin note: text overlap with arXiv:1503.08155 | null | null | null | cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Traditional way of storing facts in triplets ({\it head\_entity, relation,
tail\_entity}), abbreviated as ({\it h, r, t}), makes the knowledge intuitively
displayed and easily acquired by mankind, but hardly computed or even reasoned
by AI machines. Inspired by the success in applying {\it Distributed
Representations} to AI-related fields, recent studies expect to represent each
entity and relation with a unique low-dimensional embedding, which is different
from the symbolic and atomic framework of displaying knowledge in triplets. In
this way, the knowledge computing and reasoning can be essentially facilitated
by means of a simple {\it vector calculation}, i.e. ${\bf h} + {\bf r} \approx
{\bf t}$. We thus contribute an effective model to learn better embeddings
satisfying the formula by pulling the positive tail entities ${\bf t^{+}}$ to
get together and close to {\bf h} + {\bf r} ({\it Nearest Neighbor}), and
simultaneously pushing the negatives ${\bf t^{-}}$ away from the positives
${\bf t^{+}}$ via keeping a {\it Large Margin}. We also design a corresponding
learning algorithm to efficiently find the optimal solution based on {\it
Stochastic Gradient Descent} in iterative fashion. Quantitative experiments
illustrate that our approach can achieve the state-of-the-art performance,
compared with several latest methods on some benchmark datasets for two
classical applications, i.e. {\it Link prediction} and {\it Triplet
classification}. Moreover, we analyze the parameter complexities among all the
evaluated models, and analytical results indicate that our model needs fewer
computational resources on outperforming the other methods.
| [
{
"version": "v1",
"created": "Tue, 7 Apr 2015 17:50:31 GMT"
}
] | 2015-04-08T00:00:00 | [
[
"Fan",
"Miao",
""
],
[
"Zhou",
"Qiang",
""
],
[
"Zheng",
"Thomas Fang",
""
],
[
"Grishman",
"Ralph",
""
]
] | TITLE: Large Margin Nearest Neighbor Embedding for Knowledge Representation
ABSTRACT: Traditional way of storing facts in triplets ({\it head\_entity, relation,
tail\_entity}), abbreviated as ({\it h, r, t}), makes the knowledge intuitively
displayed and easily acquired by mankind, but hardly computed or even reasoned
by AI machines. Inspired by the success in applying {\it Distributed
Representations} to AI-related fields, recent studies expect to represent each
entity and relation with a unique low-dimensional embedding, which is different
from the symbolic and atomic framework of displaying knowledge in triplets. In
this way, the knowledge computing and reasoning can be essentially facilitated
by means of a simple {\it vector calculation}, i.e. ${\bf h} + {\bf r} \approx
{\bf t}$. We thus contribute an effective model to learn better embeddings
satisfying the formula by pulling the positive tail entities ${\bf t^{+}}$ to
get together and close to {\bf h} + {\bf r} ({\it Nearest Neighbor}), and
simultaneously pushing the negatives ${\bf t^{-}}$ away from the positives
${\bf t^{+}}$ via keeping a {\it Large Margin}. We also design a corresponding
learning algorithm to efficiently find the optimal solution based on {\it
Stochastic Gradient Descent} in iterative fashion. Quantitative experiments
illustrate that our approach can achieve the state-of-the-art performance,
compared with several latest methods on some benchmark datasets for two
classical applications, i.e. {\it Link prediction} and {\it Triplet
classification}. Moreover, we analyze the parameter complexities among all the
evaluated models, and analytical results indicate that our model needs fewer
computational resources on outperforming the other methods.
| no_new_dataset | 0.949623 |
1504.01697 | Alex Gittens | Jiyan Yang and Alex Gittens | Tensor machines for learning target-specific polynomial features | 19 pages, 4 color figures, 2 tables. Submitted to ECML 2015 | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent years have demonstrated that using random feature maps can
significantly decrease the training and testing times of kernel-based
algorithms without significantly lowering their accuracy. Regrettably, because
random features are target-agnostic, typically thousands of such features are
necessary to achieve acceptable accuracies. In this work, we consider the
problem of learning a small number of explicit polynomial features. Our
approach, named Tensor Machines, finds a parsimonious set of features by
optimizing over the hypothesis class introduced by Kar and Karnick for random
feature maps in a target-specific manner. Exploiting a natural connection
between polynomials and tensors, we provide bounds on the generalization error
of Tensor Machines. Empirically, Tensor Machines behave favorably on several
real-world datasets compared to other state-of-the-art techniques for learning
polynomial features, and deliver significantly more parsimonious models.
| [
{
"version": "v1",
"created": "Tue, 7 Apr 2015 18:21:37 GMT"
}
] | 2015-04-08T00:00:00 | [
[
"Yang",
"Jiyan",
""
],
[
"Gittens",
"Alex",
""
]
] | TITLE: Tensor machines for learning target-specific polynomial features
ABSTRACT: Recent years have demonstrated that using random feature maps can
significantly decrease the training and testing times of kernel-based
algorithms without significantly lowering their accuracy. Regrettably, because
random features are target-agnostic, typically thousands of such features are
necessary to achieve acceptable accuracies. In this work, we consider the
problem of learning a small number of explicit polynomial features. Our
approach, named Tensor Machines, finds a parsimonious set of features by
optimizing over the hypothesis class introduced by Kar and Karnick for random
feature maps in a target-specific manner. Exploiting a natural connection
between polynomials and tensors, we provide bounds on the generalization error
of Tensor Machines. Empirically, Tensor Machines behave favorably on several
real-world datasets compared to other state-of-the-art techniques for learning
polynomial features, and deliver significantly more parsimonious models.
| no_new_dataset | 0.947381 |
1411.1455 | Saravanan Thirumuruganathan | Md Farhadur Rahman, Weimo Liu, Saravanan Thirumuruganathan, Nan Zhang,
Gautam Das | Rank-Based Inference over Web Databases | null | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, there has been much research in Ranked Retrieval model in
structured databases, especially those in web databases. With this model, a
search query returns top-k tuples according to not just exact matches of
selection conditions, but a suitable ranking function. This paper studies a
novel problem on the privacy implications of database ranking. The motivation
is a novel yet serious privacy leakage we found on real-world web databases
which is caused by the ranking function design. Many such databases feature
private attributes - e.g., a social network allows users to specify certain
attributes as only visible to him/herself, but not to others. While these
websites generally respect the privacy settings by not directly displaying
private attribute values in search query answers, many of them nevertheless
take into account such private attributes in the ranking function design. The
conventional belief might be that tuple ranks alone are not enough to reveal
the private attribute values. Our investigation, however, shows that this is
not the case in reality.
To address the problem, we introduce a taxonomy of the problem space with two
dimensions, (1) the type of query interface and (2) the capability of
adversaries. For each subspace, we develop a novel technique which either
guarantees the successful inference of private attributes, or does so for a
significant portion of real-world tuples. We demonstrate the effectiveness and
efficiency of our techniques through theoretical analysis, extensive
experiments over real-world datasets, as well as successful online attacks over
websites with tens to hundreds of millions of users - e.g., Amazon Goodreads
and Renren.com.
| [
{
"version": "v1",
"created": "Thu, 6 Nov 2014 00:06:44 GMT"
},
{
"version": "v2",
"created": "Tue, 11 Nov 2014 00:36:10 GMT"
},
{
"version": "v3",
"created": "Mon, 2 Feb 2015 07:37:56 GMT"
},
{
"version": "v4",
"created": "Mon, 6 Apr 2015 02:03:27 GMT"
}
] | 2015-04-07T00:00:00 | [
[
"Rahman",
"Md Farhadur",
""
],
[
"Liu",
"Weimo",
""
],
[
"Thirumuruganathan",
"Saravanan",
""
],
[
"Zhang",
"Nan",
""
],
[
"Das",
"Gautam",
""
]
] | TITLE: Rank-Based Inference over Web Databases
ABSTRACT: In recent years, there has been much research in Ranked Retrieval model in
structured databases, especially those in web databases. With this model, a
search query returns top-k tuples according to not just exact matches of
selection conditions, but a suitable ranking function. This paper studies a
novel problem on the privacy implications of database ranking. The motivation
is a novel yet serious privacy leakage we found on real-world web databases
which is caused by the ranking function design. Many such databases feature
private attributes - e.g., a social network allows users to specify certain
attributes as only visible to him/herself, but not to others. While these
websites generally respect the privacy settings by not directly displaying
private attribute values in search query answers, many of them nevertheless
take into account such private attributes in the ranking function design. The
conventional belief might be that tuple ranks alone are not enough to reveal
the private attribute values. Our investigation, however, shows that this is
not the case in reality.
To address the problem, we introduce a taxonomy of the problem space with two
dimensions, (1) the type of query interface and (2) the capability of
adversaries. For each subspace, we develop a novel technique which either
guarantees the successful inference of private attributes, or does so for a
significant portion of real-world tuples. We demonstrate the effectiveness and
efficiency of our techniques through theoretical analysis, extensive
experiments over real-world datasets, as well as successful online attacks over
websites with tens to hundreds of millions of users - e.g., Amazon Goodreads
and Renren.com.
| no_new_dataset | 0.946101 |
1504.00325 | C. Lawrence Zitnick | Xinlei Chen, Hao Fang, Tsung-Yi Lin, Ramakrishna Vedantam, Saurabh
Gupta, Piotr Dollar, C. Lawrence Zitnick | Microsoft COCO Captions: Data Collection and Evaluation Server | arXiv admin note: text overlap with arXiv:1411.4952 | null | null | null | cs.CV cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we describe the Microsoft COCO Caption dataset and evaluation
server. When completed, the dataset will contain over one and a half million
captions describing over 330,000 images. For the training and validation
images, five independent human generated captions will be provided. To ensure
consistency in evaluation of automatic caption generation algorithms, an
evaluation server is used. The evaluation server receives candidate captions
and scores them using several popular metrics, including BLEU, METEOR, ROUGE
and CIDEr. Instructions for using the evaluation server are provided.
| [
{
"version": "v1",
"created": "Wed, 1 Apr 2015 18:13:43 GMT"
},
{
"version": "v2",
"created": "Fri, 3 Apr 2015 20:21:16 GMT"
}
] | 2015-04-07T00:00:00 | [
[
"Chen",
"Xinlei",
""
],
[
"Fang",
"Hao",
""
],
[
"Lin",
"Tsung-Yi",
""
],
[
"Vedantam",
"Ramakrishna",
""
],
[
"Gupta",
"Saurabh",
""
],
[
"Dollar",
"Piotr",
""
],
[
"Zitnick",
"C. Lawrence",
""
]
] | TITLE: Microsoft COCO Captions: Data Collection and Evaluation Server
ABSTRACT: In this paper we describe the Microsoft COCO Caption dataset and evaluation
server. When completed, the dataset will contain over one and a half million
captions describing over 330,000 images. For the training and validation
images, five independent human generated captions will be provided. To ensure
consistency in evaluation of automatic caption generation algorithms, an
evaluation server is used. The evaluation server receives candidate captions
and scores them using several popular metrics, including BLEU, METEOR, ROUGE
and CIDEr. Instructions for using the evaluation server are provided.
| no_new_dataset | 0.939582 |
1504.01142 | Nam-phuong Nguyen | Nam-phuong Nguyen, Siavash Mirarab, Keerthana Kumar, Tandy Warnow | Ultra-large alignments using Phylogeny-aware Profiles | Online supplemental materials and data are available at
http://www.cs.utexas.edu/users/phylo/software/upp/ | null | null | null | q-bio.GN cs.CE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many biological questions, including the estimation of deep evolutionary
histories and the detection of remote homology between protein sequences, rely
upon multiple sequence alignments (MSAs) and phylogenetic trees of large
datasets. However, accurate large-scale multiple sequence alignment is very
difficult, especially when the dataset contains fragmentary sequences. We
present UPP, an MSA method that uses a new machine learning technique - the
Ensemble of Hidden Markov Models - that we propose here. UPP produces highly
accurate alignments for both nucleotide and amino acid sequences, even on
ultra-large datasets or datasets containing fragmentary sequences. UPP is
available at https://github.com/smirarab/sepp.
| [
{
"version": "v1",
"created": "Sun, 5 Apr 2015 17:15:38 GMT"
}
] | 2015-04-07T00:00:00 | [
[
"Nguyen",
"Nam-phuong",
""
],
[
"Mirarab",
"Siavash",
""
],
[
"Kumar",
"Keerthana",
""
],
[
"Warnow",
"Tandy",
""
]
] | TITLE: Ultra-large alignments using Phylogeny-aware Profiles
ABSTRACT: Many biological questions, including the estimation of deep evolutionary
histories and the detection of remote homology between protein sequences, rely
upon multiple sequence alignments (MSAs) and phylogenetic trees of large
datasets. However, accurate large-scale multiple sequence alignment is very
difficult, especially when the dataset contains fragmentary sequences. We
present UPP, an MSA method that uses a new machine learning technique - the
Ensemble of Hidden Markov Models - that we propose here. UPP produces highly
accurate alignments for both nucleotide and amino acid sequences, even on
ultra-large datasets or datasets containing fragmentary sequences. UPP is
available at https://github.com/smirarab/sepp.
| no_new_dataset | 0.951997 |
1504.01220 | Xiaodan Liang | Si Liu and Xiaodan Liang and Luoqi Liu and Xiaohui Shen and Jianchao
Yang and Changsheng Xu and Liang Lin and Xiaochun Cao and Shuicheng Yan | Matching-CNN Meets KNN: Quasi-Parametric Human Parsing | This manuscript is the accepted version for CVPR 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Both parametric and non-parametric approaches have demonstrated encouraging
performances in the human parsing task, namely segmenting a human image into
several semantic regions (e.g., hat, bag, left arm, face). In this work, we aim
to develop a new solution with the advantages of both methodologies, namely
supervision from annotated data and the flexibility to use newly annotated
(possibly uncommon) images, and present a quasi-parametric human parsing model.
Under the classic K Nearest Neighbor (KNN)-based nonparametric framework, the
parametric Matching Convolutional Neural Network (M-CNN) is proposed to predict
the matching confidence and displacements of the best matched region in the
testing image for a particular semantic region in one KNN image. Given a
testing image, we first retrieve its KNN images from the
annotated/manually-parsed human image corpus. Then each semantic region in each
KNN image is matched with confidence to the testing image using M-CNN, and the
matched regions from all KNN images are further fused, followed by a superpixel
smoothing procedure to obtain the ultimate human parsing result. The M-CNN
differs from the classic CNN in that the tailored cross image matching filters
are introduced to characterize the matching between the testing image and the
semantic region of a KNN image. The cross image matching filters are defined at
different convolutional layers, each aiming to capture a particular range of
displacements. Comprehensive evaluations over a large dataset with 7,700
annotated human images well demonstrate the significant performance gain from
the quasi-parametric model over the state-of-the-arts, for the human parsing
task.
| [
{
"version": "v1",
"created": "Mon, 6 Apr 2015 07:20:02 GMT"
}
] | 2015-04-07T00:00:00 | [
[
"Liu",
"Si",
""
],
[
"Liang",
"Xiaodan",
""
],
[
"Liu",
"Luoqi",
""
],
[
"Shen",
"Xiaohui",
""
],
[
"Yang",
"Jianchao",
""
],
[
"Xu",
"Changsheng",
""
],
[
"Lin",
"Liang",
""
],
[
"Cao",
"Xiaochun",
""
],
[
"Yan",
"Shuicheng",
""
]
] | TITLE: Matching-CNN Meets KNN: Quasi-Parametric Human Parsing
ABSTRACT: Both parametric and non-parametric approaches have demonstrated encouraging
performances in the human parsing task, namely segmenting a human image into
several semantic regions (e.g., hat, bag, left arm, face). In this work, we aim
to develop a new solution with the advantages of both methodologies, namely
supervision from annotated data and the flexibility to use newly annotated
(possibly uncommon) images, and present a quasi-parametric human parsing model.
Under the classic K Nearest Neighbor (KNN)-based nonparametric framework, the
parametric Matching Convolutional Neural Network (M-CNN) is proposed to predict
the matching confidence and displacements of the best matched region in the
testing image for a particular semantic region in one KNN image. Given a
testing image, we first retrieve its KNN images from the
annotated/manually-parsed human image corpus. Then each semantic region in each
KNN image is matched with confidence to the testing image using M-CNN, and the
matched regions from all KNN images are further fused, followed by a superpixel
smoothing procedure to obtain the ultimate human parsing result. The M-CNN
differs from the classic CNN in that the tailored cross image matching filters
are introduced to characterize the matching between the testing image and the
semantic region of a KNN image. The cross image matching filters are defined at
different convolutional layers, each aiming to capture a particular range of
displacements. Comprehensive evaluations over a large dataset with 7,700
annotated human images well demonstrate the significant performance gain from
the quasi-parametric model over the state-of-the-arts, for the human parsing
task.
| no_new_dataset | 0.951097 |
1504.01304 | Yuzhen Ye | Yuzhen Ye and Haixu Tang | Utilizing de Bruijn graph of metagenome assembly for metatranscriptome
analysis | 8 pages, 4 figures, accepted in RECOMB-Seq 2015, under consideration
in Bioinformatics (a special issue for RECOMB-Seq/CBB) | null | null | null | q-bio.GN cs.CE cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Metagenomics research has accelerated the studies of microbial organisms,
providing insights into the composition and potential functionality of various
microbial communities. Metatranscriptomics (studies of the transcripts from a
mixture of microbial species) and other meta-omics approaches hold even greater
promise for providing additional insights into functional and regulatory
characteristics of the microbial communities. Current metatranscriptomics
projects are often carried out without matched metagenomic datasets (of the
same microbial communities). For the projects that produce both
metatranscriptomic and metagenomic datasets, their analyses are often not
integrated. Metagenome assemblies are far from perfect, partially explaining
why metagenome assemblies are not used for the analysis of metatranscriptomic
datasets. Here we report a reads mapping algorithm for mapping of short reads
onto a de Bruijn graph of assemblies. A hash table of junction k-mers (k-mers
spanning branching structures in the de Bruijn graph) is used to facilitate
fast mapping of reads to the graph. We developed an application of this mapping
algorithm: a reference based approach to metatranscriptome assembly using
graphs of metagenome assembly as the reference. Our results show that this new
approach (called TAG) helps to assemble substantially more transcripts that
otherwise would have been missed or truncated because of the fragmented nature
of the reference metagenome. TAG was implemented in C++ and has been tested
extensively on the linux platform. It is available for download as open source
at http://omics.informatics.indiana.edu/TAG.
| [
{
"version": "v1",
"created": "Mon, 6 Apr 2015 15:17:29 GMT"
}
] | 2015-04-07T00:00:00 | [
[
"Ye",
"Yuzhen",
""
],
[
"Tang",
"Haixu",
""
]
] | TITLE: Utilizing de Bruijn graph of metagenome assembly for metatranscriptome
analysis
ABSTRACT: Metagenomics research has accelerated the studies of microbial organisms,
providing insights into the composition and potential functionality of various
microbial communities. Metatranscriptomics (studies of the transcripts from a
mixture of microbial species) and other meta-omics approaches hold even greater
promise for providing additional insights into functional and regulatory
characteristics of the microbial communities. Current metatranscriptomics
projects are often carried out without matched metagenomic datasets (of the
same microbial communities). For the projects that produce both
metatranscriptomic and metagenomic datasets, their analyses are often not
integrated. Metagenome assemblies are far from perfect, partially explaining
why metagenome assemblies are not used for the analysis of metatranscriptomic
datasets. Here we report a reads mapping algorithm for mapping of short reads
onto a de Bruijn graph of assemblies. A hash table of junction k-mers (k-mers
spanning branching structures in the de Bruijn graph) is used to facilitate
fast mapping of reads to the graph. We developed an application of this mapping
algorithm: a reference based approach to metatranscriptome assembly using
graphs of metagenome assembly as the reference. Our results show that this new
approach (called TAG) helps to assemble substantially more transcripts that
otherwise would have been missed or truncated because of the fragmented nature
of the reference metagenome. TAG was implemented in C++ and has been tested
extensively on the linux platform. It is available for download as open source
at http://omics.informatics.indiana.edu/TAG.
| no_new_dataset | 0.944382 |
1412.1897 | Anh Nguyen | Anh Nguyen, Jason Yosinski, Jeff Clune | Deep Neural Networks are Easily Fooled: High Confidence Predictions for
Unrecognizable Images | To appear at CVPR 2015 | null | null | null | cs.CV cs.AI cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep neural networks (DNNs) have recently been achieving state-of-the-art
performance on a variety of pattern-recognition tasks, most notably visual
classification problems. Given that DNNs are now able to classify objects in
images with near-human-level performance, questions naturally arise as to what
differences remain between computer and human vision. A recent study revealed
that changing an image (e.g. of a lion) in a way imperceptible to humans can
cause a DNN to label the image as something else entirely (e.g. mislabeling a
lion a library). Here we show a related result: it is easy to produce images
that are completely unrecognizable to humans, but that state-of-the-art DNNs
believe to be recognizable objects with 99.99% confidence (e.g. labeling with
certainty that white noise static is a lion). Specifically, we take
convolutional neural networks trained to perform well on either the ImageNet or
MNIST datasets and then find images with evolutionary algorithms or gradient
ascent that DNNs label with high confidence as belonging to each dataset class.
It is possible to produce images totally unrecognizable to human eyes that DNNs
believe with near certainty are familiar objects, which we call "fooling
images" (more generally, fooling examples). Our results shed light on
interesting differences between human vision and current DNNs, and raise
questions about the generality of DNN computer vision.
| [
{
"version": "v1",
"created": "Fri, 5 Dec 2014 05:29:43 GMT"
},
{
"version": "v2",
"created": "Thu, 18 Dec 2014 22:27:00 GMT"
},
{
"version": "v3",
"created": "Tue, 3 Mar 2015 16:41:04 GMT"
},
{
"version": "v4",
"created": "Thu, 2 Apr 2015 23:12:56 GMT"
}
] | 2015-04-06T00:00:00 | [
[
"Nguyen",
"Anh",
""
],
[
"Yosinski",
"Jason",
""
],
[
"Clune",
"Jeff",
""
]
] | TITLE: Deep Neural Networks are Easily Fooled: High Confidence Predictions for
Unrecognizable Images
ABSTRACT: Deep neural networks (DNNs) have recently been achieving state-of-the-art
performance on a variety of pattern-recognition tasks, most notably visual
classification problems. Given that DNNs are now able to classify objects in
images with near-human-level performance, questions naturally arise as to what
differences remain between computer and human vision. A recent study revealed
that changing an image (e.g. of a lion) in a way imperceptible to humans can
cause a DNN to label the image as something else entirely (e.g. mislabeling a
lion a library). Here we show a related result: it is easy to produce images
that are completely unrecognizable to humans, but that state-of-the-art DNNs
believe to be recognizable objects with 99.99% confidence (e.g. labeling with
certainty that white noise static is a lion). Specifically, we take
convolutional neural networks trained to perform well on either the ImageNet or
MNIST datasets and then find images with evolutionary algorithms or gradient
ascent that DNNs label with high confidence as belonging to each dataset class.
It is possible to produce images totally unrecognizable to human eyes that DNNs
believe with near certainty are familiar objects, which we call "fooling
images" (more generally, fooling examples). Our results shed light on
interesting differences between human vision and current DNNs, and raise
questions about the generality of DNN computer vision.
| no_new_dataset | 0.939692 |
1412.5758 | Zhangyang Wang | Zhangyang Wang, Jianchao Yang, Hailin Jin, Eli Shechtman, Aseem
Agarwala, Jonathan Brandt, Thomas S. Huang | Decomposition-Based Domain Adaptation for Real-World Font Recognition | This paper has been withdrawn by the author due to project concerns | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a domain adaption framework to address a domain mismatch between
synthetic training and real-world testing data. We demonstrate our method on a
challenging fine-grain classification problem: recognizing a font style from an
image of text. In this task, it is very easy to generate lots of rendered font
examples but very hard to obtain real-world labeled images. This
real-to-synthetic domain gap caused poor generalization to new real data in
previous font recognition methods (Chen et al. (2014)). In this paper, we
introduce a Convolutional Neural Network decomposition approach, leveraging a
large training corpus of synthetic data to obtain effective features for
classification. This is done using an adaptation technique based on a Stacked
Convolutional Auto-Encoder that exploits a large collection of unlabeled
real-world text images combined with synthetic data preprocessed in a specific
way. The proposed DeepFont method achieves an accuracy of higher than 80%
(top-5) on a new large labeled real-world dataset we collected.
| [
{
"version": "v1",
"created": "Thu, 18 Dec 2014 08:51:15 GMT"
},
{
"version": "v2",
"created": "Fri, 19 Dec 2014 02:45:35 GMT"
},
{
"version": "v3",
"created": "Sun, 25 Jan 2015 22:41:40 GMT"
},
{
"version": "v4",
"created": "Wed, 1 Apr 2015 22:02:59 GMT"
}
] | 2015-04-03T00:00:00 | [
[
"Wang",
"Zhangyang",
""
],
[
"Yang",
"Jianchao",
""
],
[
"Jin",
"Hailin",
""
],
[
"Shechtman",
"Eli",
""
],
[
"Agarwala",
"Aseem",
""
],
[
"Brandt",
"Jonathan",
""
],
[
"Huang",
"Thomas S.",
""
]
] | TITLE: Decomposition-Based Domain Adaptation for Real-World Font Recognition
ABSTRACT: We present a domain adaption framework to address a domain mismatch between
synthetic training and real-world testing data. We demonstrate our method on a
challenging fine-grain classification problem: recognizing a font style from an
image of text. In this task, it is very easy to generate lots of rendered font
examples but very hard to obtain real-world labeled images. This
real-to-synthetic domain gap caused poor generalization to new real data in
previous font recognition methods (Chen et al. (2014)). In this paper, we
introduce a Convolutional Neural Network decomposition approach, leveraging a
large training corpus of synthetic data to obtain effective features for
classification. This is done using an adaptation technique based on a Stacked
Convolutional Auto-Encoder that exploits a large collection of unlabeled
real-world text images combined with synthetic data preprocessed in a specific
way. The proposed DeepFont method achieves an accuracy of higher than 80%
(top-5) on a new large labeled real-world dataset we collected.
| new_dataset | 0.950686 |
1502.06306 | Jinseok Kim | Jinseok Kim and Jana Diesner | Distortive Effects of Initial-Based Name Disambiguation on Measurements
of Large-Scale Coauthorship Networks | This is a preprint of an article accepted for publication in Journal
of the Association for Information Science and Technology | null | 10.1002/asi.23489 | null | cs.DL cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Scholars have often relied on name initials to resolve name ambiguities in
large-scale coauthorship network research. This approach bears the risk of
incorrectly merging or splitting author identities. The use of initial-based
disambiguation has been justified by the assumption that such errors would not
affect research findings too much. This paper tests this assumption by
analyzing coauthorship networks from five academic fields - biology, computer
science, nanoscience, neuroscience, and physics - and an interdisciplinary
journal, PNAS. Name instances in datasets of this study were disambiguated
based on heuristics gained from previous algorithmic disambiguation solutions.
We use disambiguated data as a proxy of ground-truth to test the performance of
three types of initial-based disambiguation. Our results show that
initial-based disambiguation can misrepresent statistical properties of
coauthorship networks: it deflates the number of unique authors, number of
component, average shortest paths, clustering coefficient, and assortativity,
while it inflates average productivity, density, average coauthor number per
author, and largest component size. Also, on average, more than half of top 10
productive or collaborative authors drop off the lists. Asian names were found
to account for the majority of misidentification by initial-based
disambiguation due to their common surname and given name initials.
| [
{
"version": "v1",
"created": "Mon, 23 Feb 2015 02:56:05 GMT"
}
] | 2015-04-03T00:00:00 | [
[
"Kim",
"Jinseok",
""
],
[
"Diesner",
"Jana",
""
]
] | TITLE: Distortive Effects of Initial-Based Name Disambiguation on Measurements
of Large-Scale Coauthorship Networks
ABSTRACT: Scholars have often relied on name initials to resolve name ambiguities in
large-scale coauthorship network research. This approach bears the risk of
incorrectly merging or splitting author identities. The use of initial-based
disambiguation has been justified by the assumption that such errors would not
affect research findings too much. This paper tests this assumption by
analyzing coauthorship networks from five academic fields - biology, computer
science, nanoscience, neuroscience, and physics - and an interdisciplinary
journal, PNAS. Name instances in datasets of this study were disambiguated
based on heuristics gained from previous algorithmic disambiguation solutions.
We use disambiguated data as a proxy of ground-truth to test the performance of
three types of initial-based disambiguation. Our results show that
initial-based disambiguation can misrepresent statistical properties of
coauthorship networks: it deflates the number of unique authors, number of
component, average shortest paths, clustering coefficient, and assortativity,
while it inflates average productivity, density, average coauthor number per
author, and largest component size. Also, on average, more than half of top 10
productive or collaborative authors drop off the lists. Asian names were found
to account for the majority of misidentification by initial-based
disambiguation due to their common surname and given name initials.
| no_new_dataset | 0.950227 |
1504.00430 | Hanyang Peng | Hanyang Peng, Yong Fan | Direct l_(2,p)-Norm Learning for Feature Selection | null | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose a novel sparse learning based feature selection
method that directly optimizes a large margin linear classification model
sparsity with l_(2,p)-norm (0 < p < 1)subject to data-fitting constraints,
rather than using the sparsity as a regularization term. To solve the direct
sparsity optimization problem that is non-smooth and non-convex when 0<p<1, we
provide an efficient iterative algorithm with proved convergence by converting
it to a convex and smooth optimization problem at every iteration step. The
proposed algorithm has been evaluated based on publicly available datasets, and
extensive comparison experiments have demonstrated that our algorithm could
achieve feature selection performance competitive to state-of-the-art
algorithms.
| [
{
"version": "v1",
"created": "Thu, 2 Apr 2015 02:16:39 GMT"
}
] | 2015-04-03T00:00:00 | [
[
"Peng",
"Hanyang",
""
],
[
"Fan",
"Yong",
""
]
] | TITLE: Direct l_(2,p)-Norm Learning for Feature Selection
ABSTRACT: In this paper, we propose a novel sparse learning based feature selection
method that directly optimizes a large margin linear classification model
sparsity with l_(2,p)-norm (0 < p < 1)subject to data-fitting constraints,
rather than using the sparsity as a regularization term. To solve the direct
sparsity optimization problem that is non-smooth and non-convex when 0<p<1, we
provide an efficient iterative algorithm with proved convergence by converting
it to a convex and smooth optimization problem at every iteration step. The
proposed algorithm has been evaluated based on publicly available datasets, and
extensive comparison experiments have demonstrated that our algorithm could
achieve feature selection performance competitive to state-of-the-art
algorithms.
| no_new_dataset | 0.949809 |
1404.0284 | Jack Kelly | Jack Kelly and William Knottenbelt | The UK-DALE dataset, domestic appliance-level electricity demand and
whole-house demand from five UK homes | null | Scientific Data 2 (2015) Article number: 150007 (2015) | 10.1038/sdata.2015.7 | null | cs.OH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many countries are rolling out smart electricity meters. These measure a
home's total power demand. However, research into consumer behaviour suggests
that consumers are best able to improve their energy efficiency when provided
with itemised, appliance-by-appliance consumption information. Energy
disaggregation is a computational technique for estimating
appliance-by-appliance energy consumption from a whole-house meter signal.
To conduct research on disaggregation algorithms, researchers require data
describing not just the aggregate demand per building but also the `ground
truth' demand of individual appliances. In this context, we present UK-DALE: an
open-access dataset from the UK recording Domestic Appliance-Level Electricity
at a sample rate of 16 kHz for the whole-house and at 1/6 Hz for individual
appliances. This is the first open access UK dataset at this temporal
resolution. We recorded from five houses, one of which was recorded for 655
days, the longest duration we are aware of for any energy dataset at this
sample rate. We also describe the low-cost, open-source, wireless system we
built for collecting our dataset.
| [
{
"version": "v1",
"created": "Tue, 1 Apr 2014 15:49:00 GMT"
},
{
"version": "v2",
"created": "Mon, 12 Jan 2015 14:29:00 GMT"
},
{
"version": "v3",
"created": "Thu, 19 Mar 2015 10:45:57 GMT"
}
] | 2015-04-02T00:00:00 | [
[
"Kelly",
"Jack",
""
],
[
"Knottenbelt",
"William",
""
]
] | TITLE: The UK-DALE dataset, domestic appliance-level electricity demand and
whole-house demand from five UK homes
ABSTRACT: Many countries are rolling out smart electricity meters. These measure a
home's total power demand. However, research into consumer behaviour suggests
that consumers are best able to improve their energy efficiency when provided
with itemised, appliance-by-appliance consumption information. Energy
disaggregation is a computational technique for estimating
appliance-by-appliance energy consumption from a whole-house meter signal.
To conduct research on disaggregation algorithms, researchers require data
describing not just the aggregate demand per building but also the `ground
truth' demand of individual appliances. In this context, we present UK-DALE: an
open-access dataset from the UK recording Domestic Appliance-Level Electricity
at a sample rate of 16 kHz for the whole-house and at 1/6 Hz for individual
appliances. This is the first open access UK dataset at this temporal
resolution. We recorded from five houses, one of which was recorded for 655
days, the longest duration we are aware of for any energy dataset at this
sample rate. We also describe the low-cost, open-source, wireless system we
built for collecting our dataset.
| new_dataset | 0.963643 |
1503.02445 | Muhammad Uzair | Muhammad Uzair, Faisal Shafait, Bernard Ghanem, Ajmal Mian | Representation Learning with Deep Extreme Learning Machines for
Efficient Image Set Classification | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Efficient and accurate joint representation of a collection of images, that
belong to the same class, is a major research challenge for practical image set
classification. Existing methods either make prior assumptions about the data
structure, or perform heavy computations to learn structure from the data
itself. In this paper, we propose an efficient image set representation that
does not make any prior assumptions about the structure of the underlying data.
We learn the non-linear structure of image sets with Deep Extreme Learning
Machines (DELM) that are very efficient and generalize well even on a limited
number of training samples. Extensive experiments on a broad range of public
datasets for image set classification (Honda/UCSD, CMU Mobo, YouTube
Celebrities, Celebrity-1000, ETH-80) show that the proposed algorithm
consistently outperforms state-of-the-art image set classification methods both
in terms of speed and accuracy.
| [
{
"version": "v1",
"created": "Mon, 9 Mar 2015 12:14:42 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Mar 2015 05:29:31 GMT"
},
{
"version": "v3",
"created": "Wed, 1 Apr 2015 10:29:09 GMT"
}
] | 2015-04-02T00:00:00 | [
[
"Uzair",
"Muhammad",
""
],
[
"Shafait",
"Faisal",
""
],
[
"Ghanem",
"Bernard",
""
],
[
"Mian",
"Ajmal",
""
]
] | TITLE: Representation Learning with Deep Extreme Learning Machines for
Efficient Image Set Classification
ABSTRACT: Efficient and accurate joint representation of a collection of images, that
belong to the same class, is a major research challenge for practical image set
classification. Existing methods either make prior assumptions about the data
structure, or perform heavy computations to learn structure from the data
itself. In this paper, we propose an efficient image set representation that
does not make any prior assumptions about the structure of the underlying data.
We learn the non-linear structure of image sets with Deep Extreme Learning
Machines (DELM) that are very efficient and generalize well even on a limited
number of training samples. Extensive experiments on a broad range of public
datasets for image set classification (Honda/UCSD, CMU Mobo, YouTube
Celebrities, Celebrity-1000, ETH-80) show that the proposed algorithm
consistently outperforms state-of-the-art image set classification methods both
in terms of speed and accuracy.
| no_new_dataset | 0.950915 |
1504.00028 | Zhangyang Wang | Zhangyang Wang, Jianchao Yang, Hailin Jin, Eli Shechtman, Aseem
Agarwala, Jonathan Brandt, Thomas S. Huang | Real-World Font Recognition Using Deep Network and Domain Adaptation | null | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We address a challenging fine-grain classification problem: recognizing a
font style from an image of text. In this task, it is very easy to generate
lots of rendered font examples but very hard to obtain real-world labeled
images. This real-to-synthetic domain gap caused poor generalization to new
real data in previous methods (Chen et al. (2014)). In this paper, we refer to
Convolutional Neural Networks, and use an adaptation technique based on a
Stacked Convolutional Auto-Encoder that exploits unlabeled real-world images
combined with synthetic data. The proposed method achieves an accuracy of
higher than 80% (top-5) on a real-world dataset.
| [
{
"version": "v1",
"created": "Tue, 31 Mar 2015 20:30:00 GMT"
}
] | 2015-04-02T00:00:00 | [
[
"Wang",
"Zhangyang",
""
],
[
"Yang",
"Jianchao",
""
],
[
"Jin",
"Hailin",
""
],
[
"Shechtman",
"Eli",
""
],
[
"Agarwala",
"Aseem",
""
],
[
"Brandt",
"Jonathan",
""
],
[
"Huang",
"Thomas S.",
""
]
] | TITLE: Real-World Font Recognition Using Deep Network and Domain Adaptation
ABSTRACT: We address a challenging fine-grain classification problem: recognizing a
font style from an image of text. In this task, it is very easy to generate
lots of rendered font examples but very hard to obtain real-world labeled
images. This real-to-synthetic domain gap caused poor generalization to new
real data in previous methods (Chen et al. (2014)). In this paper, we refer to
Convolutional Neural Networks, and use an adaptation technique based on a
Stacked Convolutional Auto-Encoder that exploits unlabeled real-world images
combined with synthetic data. The proposed method achieves an accuracy of
higher than 80% (top-5) on a real-world dataset.
| no_new_dataset | 0.954774 |
1504.00045 | Yongxin Yang | Zhiyuan Shi, Yongxin Yang, Timothy M. Hospedales and Tao Xiang | Weakly Supervised Learning of Objects, Attributes and their Associations | 14 pages, Accepted to ECCV 2014 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When humans describe images they tend to use combinations of nouns and
adjectives, corresponding to objects and their associated attributes
respectively. To generate such a description automatically, one needs to model
objects, attributes and their associations. Conventional methods require strong
annotation of object and attribute locations, making them less scalable. In
this paper, we model object-attribute associations from weakly labelled images,
such as those widely available on media sharing sites (e.g. Flickr), where only
image-level labels (either object or attributes) are given, without their
locations and associations. This is achieved by introducing a novel weakly
supervised non-parametric Bayesian model. Once learned, given a new image, our
model can describe the image, including objects, attributes and their
associations, as well as their locations and segmentation. Extensive
experiments on benchmark datasets demonstrate that our weakly supervised model
performs at par with strongly supervised models on tasks such as image
description and retrieval based on object-attribute associations.
| [
{
"version": "v1",
"created": "Tue, 31 Mar 2015 21:18:18 GMT"
}
] | 2015-04-02T00:00:00 | [
[
"Shi",
"Zhiyuan",
""
],
[
"Yang",
"Yongxin",
""
],
[
"Hospedales",
"Timothy M.",
""
],
[
"Xiang",
"Tao",
""
]
] | TITLE: Weakly Supervised Learning of Objects, Attributes and their Associations
ABSTRACT: When humans describe images they tend to use combinations of nouns and
adjectives, corresponding to objects and their associated attributes
respectively. To generate such a description automatically, one needs to model
objects, attributes and their associations. Conventional methods require strong
annotation of object and attribute locations, making them less scalable. In
this paper, we model object-attribute associations from weakly labelled images,
such as those widely available on media sharing sites (e.g. Flickr), where only
image-level labels (either object or attributes) are given, without their
locations and associations. This is achieved by introducing a novel weakly
supervised non-parametric Bayesian model. Once learned, given a new image, our
model can describe the image, including objects, attributes and their
associations, as well as their locations and segmentation. Extensive
experiments on benchmark datasets demonstrate that our weakly supervised model
performs at par with strongly supervised models on tasks such as image
description and retrieval based on object-attribute associations.
| no_new_dataset | 0.953188 |
1504.00191 | Sanjay Sahay | Rajendra Kumar Roul, Shubham Rohan Asthana, Sanjay Kumar Sahay | Automated Document Indexing via Intelligent Hierarchical Clustering: A
Novel Approach | 6 Pages, 3 Figures. IEEE Xplore, ICHPCA-2014 | null | 10.1109/ICHPCA.2014.7045347 | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the rising quantity of textual data available in electronic format, the
need to organize it become a highly challenging task. In the present paper, we
explore a document organization framework that exploits an intelligent
hierarchical clustering algorithm to generate an index over a set of documents.
The framework has been designed to be scalable and accurate even with large
corpora. The advantage of the proposed algorithm lies in the need for minimal
inputs, with much of the hierarchy attributes being decided in an automated
manner using statistical methods. The use of topic modeling in a pre-processing
stage ensures robustness to a range of variations in the input data. For
experimental work 20-Newsgroups dataset has been used. The F- measure of the
proposed approach has been compared with the traditional K-Means and K-Medoids
clustering algorithms. Test results demonstrate the applicability, efficiency
and effectiveness of our proposed approach. After extensive experimentation, we
conclude that the framework shows promise for further research and specialized
commercial applications.
| [
{
"version": "v1",
"created": "Wed, 1 Apr 2015 12:08:36 GMT"
}
] | 2015-04-02T00:00:00 | [
[
"Roul",
"Rajendra Kumar",
""
],
[
"Asthana",
"Shubham Rohan",
""
],
[
"Sahay",
"Sanjay Kumar",
""
]
] | TITLE: Automated Document Indexing via Intelligent Hierarchical Clustering: A
Novel Approach
ABSTRACT: With the rising quantity of textual data available in electronic format, the
need to organize it become a highly challenging task. In the present paper, we
explore a document organization framework that exploits an intelligent
hierarchical clustering algorithm to generate an index over a set of documents.
The framework has been designed to be scalable and accurate even with large
corpora. The advantage of the proposed algorithm lies in the need for minimal
inputs, with much of the hierarchy attributes being decided in an automated
manner using statistical methods. The use of topic modeling in a pre-processing
stage ensures robustness to a range of variations in the input data. For
experimental work 20-Newsgroups dataset has been used. The F- measure of the
proposed approach has been compared with the traditional K-Means and K-Medoids
clustering algorithms. Test results demonstrate the applicability, efficiency
and effectiveness of our proposed approach. After extensive experimentation, we
conclude that the framework shows promise for further research and specialized
commercial applications.
| no_new_dataset | 0.949435 |
1504.00331 | Eldon Carman Jr | E. Preston Carman Jr. (1), Till Westmann (2), Vinayak R. Borkar (3),
Michael J. Carey (3) and Vassilis J. Tsotras (1) ((1) UC Riverside, (2)
Oracle Labs, (3) UC Irvine) | Apache VXQuery: A Scalable XQuery Implementation | null | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The wide use of XML for document management and data exchange has created the
need to query large repositories of XML data. To efficiently query such large
data collections and take advantage of parallelism, we have implemented Apache
VXQuery, an open-source scalable XQuery processor. The system builds upon two
other open-source frameworks -- Hyracks, a parallel execution engine, and
Algebricks, a language agnostic compiler toolbox. Apache VXQuery extends these
two frameworks and provides an implementation of the XQuery specifics (data
model, data-model dependent functions and optimizations, and a parser). We
describe the architecture of Apache VXQuery, its integration with Hyracks and
Algebricks, and the XQuery optimization rules applied to the query plan to
improve path expression efficiency and to enable query parallelism. An
experimental evaluation using a real 500GB dataset with various selection,
aggregation and join XML queries shows that Apache VXQuery performs well both
in terms of scale-up and speed-up. Our experiments show that it is about 3x
faster than Saxon (an open-source and commercial XQuery processor) on a 4-core,
single node implementation, and around 2.5x faster than Apache MRQL (a
MapReduce-based parallel query processor) on an eight (4-core) node cluster.
| [
{
"version": "v1",
"created": "Wed, 1 Apr 2015 18:27:23 GMT"
}
] | 2015-04-02T00:00:00 | [
[
"Carman",
"E. Preston",
"Jr."
],
[
"Westmann",
"Till",
""
],
[
"Borkar",
"Vinayak R.",
""
],
[
"Carey",
"Michael J.",
""
],
[
"Tsotras",
"Vassilis J.",
""
]
] | TITLE: Apache VXQuery: A Scalable XQuery Implementation
ABSTRACT: The wide use of XML for document management and data exchange has created the
need to query large repositories of XML data. To efficiently query such large
data collections and take advantage of parallelism, we have implemented Apache
VXQuery, an open-source scalable XQuery processor. The system builds upon two
other open-source frameworks -- Hyracks, a parallel execution engine, and
Algebricks, a language agnostic compiler toolbox. Apache VXQuery extends these
two frameworks and provides an implementation of the XQuery specifics (data
model, data-model dependent functions and optimizations, and a parser). We
describe the architecture of Apache VXQuery, its integration with Hyracks and
Algebricks, and the XQuery optimization rules applied to the query plan to
improve path expression efficiency and to enable query parallelism. An
experimental evaluation using a real 500GB dataset with various selection,
aggregation and join XML queries shows that Apache VXQuery performs well both
in terms of scale-up and speed-up. Our experiments show that it is about 3x
faster than Saxon (an open-source and commercial XQuery processor) on a 4-core,
single node implementation, and around 2.5x faster than Apache MRQL (a
MapReduce-based parallel query processor) on an eight (4-core) node cluster.
| no_new_dataset | 0.941761 |
1306.6802 | Aris Kosmopoulos | Aris Kosmopoulos, Ioannis Partalas, Eric Gaussier, Georgios Paliouras,
Ion Androutsopoulos | Evaluation Measures for Hierarchical Classification: a unified view and
novel approaches | Submitted to journal | null | 10.1007/s10618-014-0382-x | null | cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hierarchical classification addresses the problem of classifying items into a
hierarchy of classes. An important issue in hierarchical classification is the
evaluation of different classification algorithms, which is complicated by the
hierarchical relations among the classes. Several evaluation measures have been
proposed for hierarchical classification using the hierarchy in different ways.
This paper studies the problem of evaluation in hierarchical classification by
analyzing and abstracting the key components of the existing performance
measures. It also proposes two alternative generic views of hierarchical
evaluation and introduces two corresponding novel measures. The proposed
measures, along with the state-of-the art ones, are empirically tested on three
large datasets from the domain of text classification. The empirical results
illustrate the undesirable behavior of existing approaches and how the proposed
methods overcome most of these methods across a range of cases.
| [
{
"version": "v1",
"created": "Fri, 28 Jun 2013 11:49:53 GMT"
},
{
"version": "v2",
"created": "Mon, 1 Jul 2013 17:33:58 GMT"
}
] | 2015-04-01T00:00:00 | [
[
"Kosmopoulos",
"Aris",
""
],
[
"Partalas",
"Ioannis",
""
],
[
"Gaussier",
"Eric",
""
],
[
"Paliouras",
"Georgios",
""
],
[
"Androutsopoulos",
"Ion",
""
]
] | TITLE: Evaluation Measures for Hierarchical Classification: a unified view and
novel approaches
ABSTRACT: Hierarchical classification addresses the problem of classifying items into a
hierarchy of classes. An important issue in hierarchical classification is the
evaluation of different classification algorithms, which is complicated by the
hierarchical relations among the classes. Several evaluation measures have been
proposed for hierarchical classification using the hierarchy in different ways.
This paper studies the problem of evaluation in hierarchical classification by
analyzing and abstracting the key components of the existing performance
measures. It also proposes two alternative generic views of hierarchical
evaluation and introduces two corresponding novel measures. The proposed
measures, along with the state-of-the art ones, are empirically tested on three
large datasets from the domain of text classification. The empirical results
illustrate the undesirable behavior of existing approaches and how the proposed
methods overcome most of these methods across a range of cases.
| no_new_dataset | 0.947624 |
1503.02725 | Abhishek Sharma | Abhishek Sharma and Oncel Tuzel and David W. Jacobs | Deep Hierarchical Parsing for Semantic Segmentation | IEEE CVPR 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes a learning-based approach to scene parsing inspired by
the deep Recursive Context Propagation Network (RCPN). RCPN is a deep
feed-forward neural network that utilizes the contextual information from the
entire image, through bottom-up followed by top-down context propagation via
random binary parse trees. This improves the feature representation of every
super-pixel in the image for better classification into semantic categories. We
analyze RCPN and propose two novel contributions to further improve the model.
We first analyze the learning of RCPN parameters and discover the presence of
bypass error paths in the computation graph of RCPN that can hinder contextual
propagation. We propose to tackle this problem by including the classification
loss of the internal nodes of the random parse trees in the original RCPN loss
function. Secondly, we use an MRF on the parse tree nodes to model the
hierarchical dependency present in the output. Both modifications provide
performance boosts over the original RCPN and the new system achieves
state-of-the-art performance on Stanford Background, SIFT-Flow and Daimler
urban datasets.
| [
{
"version": "v1",
"created": "Mon, 9 Mar 2015 23:05:26 GMT"
},
{
"version": "v2",
"created": "Mon, 30 Mar 2015 20:03:01 GMT"
}
] | 2015-04-01T00:00:00 | [
[
"Sharma",
"Abhishek",
""
],
[
"Tuzel",
"Oncel",
""
],
[
"Jacobs",
"David W.",
""
]
] | TITLE: Deep Hierarchical Parsing for Semantic Segmentation
ABSTRACT: This paper proposes a learning-based approach to scene parsing inspired by
the deep Recursive Context Propagation Network (RCPN). RCPN is a deep
feed-forward neural network that utilizes the contextual information from the
entire image, through bottom-up followed by top-down context propagation via
random binary parse trees. This improves the feature representation of every
super-pixel in the image for better classification into semantic categories. We
analyze RCPN and propose two novel contributions to further improve the model.
We first analyze the learning of RCPN parameters and discover the presence of
bypass error paths in the computation graph of RCPN that can hinder contextual
propagation. We propose to tackle this problem by including the classification
loss of the internal nodes of the random parse trees in the original RCPN loss
function. Secondly, we use an MRF on the parse tree nodes to model the
hierarchical dependency present in the output. Both modifications provide
performance boosts over the original RCPN and the new system achieves
state-of-the-art performance on Stanford Background, SIFT-Flow and Daimler
urban datasets.
| no_new_dataset | 0.948489 |
1503.08843 | Peng Sun | Peng Sun, James K. Min, Guanglei Xiong | Globally Tuned Cascade Pose Regression via Back Propagation with
Application in 2D Face Pose Estimation and Heart Segmentation in 3D CT Images | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, a successful pose estimation algorithm, called Cascade Pose
Regression (CPR), was proposed in the literature. Trained over Pose Index
Feature, CPR is a regressor ensemble that is similar to Boosting. In this paper
we show how CPR can be represented as a Neural Network. Specifically, we adopt
a Graph Transformer Network (GTN) representation and accordingly train CPR with
Back Propagation (BP) that permits globally tuning. In contrast, previous CPR
literature only took a layer wise training without any post fine tuning. We
empirically show that global training with BP outperforms layer-wise
(pre-)training. Our CPR-GTN adopts a Multi Layer Percetron as the regressor,
which utilized sparse connection to learn local image feature representation.
We tested the proposed CPR-GTN on 2D face pose estimation problem as in
previous CPR literature. Besides, we also investigated the possibility of
extending CPR-GTN to 3D pose estimation by doing experiments using 3D Computed
Tomography dataset for heart segmentation.
| [
{
"version": "v1",
"created": "Mon, 30 Mar 2015 20:17:23 GMT"
}
] | 2015-04-01T00:00:00 | [
[
"Sun",
"Peng",
""
],
[
"Min",
"James K.",
""
],
[
"Xiong",
"Guanglei",
""
]
] | TITLE: Globally Tuned Cascade Pose Regression via Back Propagation with
Application in 2D Face Pose Estimation and Heart Segmentation in 3D CT Images
ABSTRACT: Recently, a successful pose estimation algorithm, called Cascade Pose
Regression (CPR), was proposed in the literature. Trained over Pose Index
Feature, CPR is a regressor ensemble that is similar to Boosting. In this paper
we show how CPR can be represented as a Neural Network. Specifically, we adopt
a Graph Transformer Network (GTN) representation and accordingly train CPR with
Back Propagation (BP) that permits globally tuning. In contrast, previous CPR
literature only took a layer wise training without any post fine tuning. We
empirically show that global training with BP outperforms layer-wise
(pre-)training. Our CPR-GTN adopts a Multi Layer Percetron as the regressor,
which utilized sparse connection to learn local image feature representation.
We tested the proposed CPR-GTN on 2D face pose estimation problem as in
previous CPR literature. Besides, we also investigated the possibility of
extending CPR-GTN to 3D pose estimation by doing experiments using 3D Computed
Tomography dataset for heart segmentation.
| no_new_dataset | 0.947624 |
1503.08853 | Ali Borji | Ali Borji and James Tanner | Reconciling saliency and object center-bias hypotheses in explaining
free-viewing fixations | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Predicting where people look in natural scenes has attracted a lot of
interest in computer vision and computational neuroscience over the past two
decades. Two seemingly contrasting categories of cues have been proposed to
influence where people look: \textit{low-level image saliency} and
\textit{high-level semantic information}. Our first contribution is to take a
detailed look at these cues to confirm the hypothesis proposed by
Henderson~\cite{henderson1993eye} and Nuthmann \&
Henderson~\cite{nuthmann2010object} that observers tend to look at the center
of objects. We analyzed fixation data for scene free-viewing over 17 observers
on 60 fully annotated images with various types of objects. Images contained
different types of scenes, such as natural scenes, line drawings, and 3D
rendered scenes. Our second contribution is to propose a simple combined model
of low-level saliency and object center-bias that outperforms each individual
component significantly over our data, as well as on the OSIE dataset by Xu et
al.~\cite{xu2014predicting}. The results reconcile saliency with object
center-bias hypotheses and highlight that both types of cues are important in
guiding fixations. Our work opens new directions to understand strategies that
humans use in observing scenes and objects, and demonstrates the construction
of combined models of low-level saliency and high-level object-based
information.
| [
{
"version": "v1",
"created": "Mon, 30 Mar 2015 21:07:53 GMT"
}
] | 2015-04-01T00:00:00 | [
[
"Borji",
"Ali",
""
],
[
"Tanner",
"James",
""
]
] | TITLE: Reconciling saliency and object center-bias hypotheses in explaining
free-viewing fixations
ABSTRACT: Predicting where people look in natural scenes has attracted a lot of
interest in computer vision and computational neuroscience over the past two
decades. Two seemingly contrasting categories of cues have been proposed to
influence where people look: \textit{low-level image saliency} and
\textit{high-level semantic information}. Our first contribution is to take a
detailed look at these cues to confirm the hypothesis proposed by
Henderson~\cite{henderson1993eye} and Nuthmann \&
Henderson~\cite{nuthmann2010object} that observers tend to look at the center
of objects. We analyzed fixation data for scene free-viewing over 17 observers
on 60 fully annotated images with various types of objects. Images contained
different types of scenes, such as natural scenes, line drawings, and 3D
rendered scenes. Our second contribution is to propose a simple combined model
of low-level saliency and object center-bias that outperforms each individual
component significantly over our data, as well as on the OSIE dataset by Xu et
al.~\cite{xu2014predicting}. The results reconcile saliency with object
center-bias hypotheses and highlight that both types of cues are important in
guiding fixations. Our work opens new directions to understand strategies that
humans use in observing scenes and objects, and demonstrates the construction
of combined models of low-level saliency and high-level object-based
information.
| no_new_dataset | 0.953579 |
1503.08873 | Paul Mineiro | Paul Mineiro and Nikos Karampatziakis | Fast Label Embeddings for Extremely Large Output Spaces | Accepted as a workshop contribution at ICLR 2015 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many modern multiclass and multilabel problems are characterized by
increasingly large output spaces. For these problems, label embeddings have
been shown to be a useful primitive that can improve computational and
statistical efficiency. In this work we utilize a correspondence between rank
constrained estimation and low dimensional label embeddings that uncovers a
fast label embedding algorithm which works in both the multiclass and
multilabel settings. The result is a randomized algorithm for partial least
squares, whose running time is exponentially faster than naive algorithms. We
demonstrate our techniques on two large-scale public datasets, from the Large
Scale Hierarchical Text Challenge and the Open Directory Project, where we
obtain state of the art results.
| [
{
"version": "v1",
"created": "Mon, 30 Mar 2015 23:29:46 GMT"
}
] | 2015-04-01T00:00:00 | [
[
"Mineiro",
"Paul",
""
],
[
"Karampatziakis",
"Nikos",
""
]
] | TITLE: Fast Label Embeddings for Extremely Large Output Spaces
ABSTRACT: Many modern multiclass and multilabel problems are characterized by
increasingly large output spaces. For these problems, label embeddings have
been shown to be a useful primitive that can improve computational and
statistical efficiency. In this work we utilize a correspondence between rank
constrained estimation and low dimensional label embeddings that uncovers a
fast label embedding algorithm which works in both the multiclass and
multilabel settings. The result is a randomized algorithm for partial least
squares, whose running time is exponentially faster than naive algorithms. We
demonstrate our techniques on two large-scale public datasets, from the Large
Scale Hierarchical Text Challenge and the Open Directory Project, where we
obtain state of the art results.
| no_new_dataset | 0.948537 |
1407.7556 | Lorenzo Livi | Lorenzo Livi, Alireza Sadeghian, Witold Pedrycz | Entropic one-class classifiers | To appear in IEEE-TNNLS | null | 10.1109/TNNLS.2015.2418332 | null | cs.CV cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The one-class classification problem is a well-known research endeavor in
pattern recognition. The problem is also known under different names, such as
outlier and novelty/anomaly detection. The core of the problem consists in
modeling and recognizing patterns belonging only to a so-called target class.
All other patterns are termed non-target, and therefore they should be
recognized as such. In this paper, we propose a novel one-class classification
system that is based on an interplay of different techniques. Primarily, we
follow a dissimilarity representation based approach; we embed the input data
into the dissimilarity space by means of an appropriate parametric
dissimilarity measure. This step allows us to process virtually any type of
data. The dissimilarity vectors are then represented through a weighted
Euclidean graphs, which we use to (i) determine the entropy of the data
distribution in the dissimilarity space, and at the same time (ii) derive
effective decision regions that are modeled as clusters of vertices. Since the
dissimilarity measure for the input data is parametric, we optimize its
parameters by means of a global optimization scheme, which considers both
mesoscopic and structural characteristics of the data represented through the
graphs. The proposed one-class classifier is designed to provide both hard
(Boolean) and soft decisions about the recognition of test patterns, allowing
an accurate description of the classification process. We evaluate the
performance of the system on different benchmarking datasets, containing either
feature-based or structured patterns. Experimental results demonstrate the
effectiveness of the proposed technique.
| [
{
"version": "v1",
"created": "Mon, 28 Jul 2014 20:26:24 GMT"
},
{
"version": "v2",
"created": "Tue, 16 Dec 2014 20:46:21 GMT"
},
{
"version": "v3",
"created": "Sun, 11 Jan 2015 16:27:23 GMT"
}
] | 2015-03-31T00:00:00 | [
[
"Livi",
"Lorenzo",
""
],
[
"Sadeghian",
"Alireza",
""
],
[
"Pedrycz",
"Witold",
""
]
] | TITLE: Entropic one-class classifiers
ABSTRACT: The one-class classification problem is a well-known research endeavor in
pattern recognition. The problem is also known under different names, such as
outlier and novelty/anomaly detection. The core of the problem consists in
modeling and recognizing patterns belonging only to a so-called target class.
All other patterns are termed non-target, and therefore they should be
recognized as such. In this paper, we propose a novel one-class classification
system that is based on an interplay of different techniques. Primarily, we
follow a dissimilarity representation based approach; we embed the input data
into the dissimilarity space by means of an appropriate parametric
dissimilarity measure. This step allows us to process virtually any type of
data. The dissimilarity vectors are then represented through a weighted
Euclidean graphs, which we use to (i) determine the entropy of the data
distribution in the dissimilarity space, and at the same time (ii) derive
effective decision regions that are modeled as clusters of vertices. Since the
dissimilarity measure for the input data is parametric, we optimize its
parameters by means of a global optimization scheme, which considers both
mesoscopic and structural characteristics of the data represented through the
graphs. The proposed one-class classifier is designed to provide both hard
(Boolean) and soft decisions about the recognition of test patterns, allowing
an accurate description of the classification process. We evaluate the
performance of the system on different benchmarking datasets, containing either
feature-based or structured patterns. Experimental results demonstrate the
effectiveness of the proposed technique.
| no_new_dataset | 0.946498 |
1410.0760 | Chih-Hang Wang | Chih-Hang Wang, De-Nian Yang, Wen-Tsuen Chen | Scheduling for Multi-Camera Surveillance in LTE Networks | 9 pages, 10 figures | null | null | null | cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Wireless surveillance in cellular networks has become increasingly important,
while commercial LTE surveillance cameras are also available nowadays.
Nevertheless, most scheduling algorithms in the literature are throughput,
fairness, or profit-based approaches, which are not suitable for wireless
surveillance. In this paper, therefore, we explore the resource allocation
problem for a multi-camera surveillance system in 3GPP Long Term Evolution
(LTE) uplink (UL) networks. We minimize the number of allocated resource blocks
(RBs) while guaranteeing the coverage requirement for surveillance systems in
LTE UL networks. Specifically, we formulate the Camera Set Resource Allocation
Problem (CSRAP) and prove that the problem is NP-Hard. We then propose an
Integer Linear Programming formulation for general cases to find the optimal
solution. Moreover, we present a baseline algorithm and devise an approximation
algorithm to solve the problem. Simulation results based on a real surveillance
map and synthetic datasets manifest that the number of allocated RBs can be
effectively reduced compared to the existing approach for LTE networks.
| [
{
"version": "v1",
"created": "Fri, 3 Oct 2014 06:17:20 GMT"
},
{
"version": "v2",
"created": "Mon, 6 Oct 2014 08:33:46 GMT"
},
{
"version": "v3",
"created": "Mon, 27 Oct 2014 04:24:11 GMT"
},
{
"version": "v4",
"created": "Sat, 28 Mar 2015 13:58:08 GMT"
}
] | 2015-03-31T00:00:00 | [
[
"Wang",
"Chih-Hang",
""
],
[
"Yang",
"De-Nian",
""
],
[
"Chen",
"Wen-Tsuen",
""
]
] | TITLE: Scheduling for Multi-Camera Surveillance in LTE Networks
ABSTRACT: Wireless surveillance in cellular networks has become increasingly important,
while commercial LTE surveillance cameras are also available nowadays.
Nevertheless, most scheduling algorithms in the literature are throughput,
fairness, or profit-based approaches, which are not suitable for wireless
surveillance. In this paper, therefore, we explore the resource allocation
problem for a multi-camera surveillance system in 3GPP Long Term Evolution
(LTE) uplink (UL) networks. We minimize the number of allocated resource blocks
(RBs) while guaranteeing the coverage requirement for surveillance systems in
LTE UL networks. Specifically, we formulate the Camera Set Resource Allocation
Problem (CSRAP) and prove that the problem is NP-Hard. We then propose an
Integer Linear Programming formulation for general cases to find the optimal
solution. Moreover, we present a baseline algorithm and devise an approximation
algorithm to solve the problem. Simulation results based on a real surveillance
map and synthetic datasets manifest that the number of allocated RBs can be
effectively reduced compared to the existing approach for LTE networks.
| no_new_dataset | 0.952926 |
1411.6400 | Min Wei | Min Wei, Tommy W. S. Chow, Rosa H. M. Chan | Mutual Information-Based Unsupervised Feature Transformation for
Heterogeneous Feature Subset Selection | This paper has been withdrawn by the author due to the number of
datasets and classifiers are not sufficient to support the claim. Need more
simulation work | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Conventional mutual information (MI) based feature selection (FS) methods are
unable to handle heterogeneous feature subset selection properly because of
data format differences or estimation methods of MI between feature subset and
class label. A way to solve this problem is feature transformation (FT). In
this study, a novel unsupervised feature transformation (UFT) which can
transform non-numerical features into numerical features is developed and
tested. The UFT process is MI-based and independent of class label. MI-based FS
algorithms, such as Parzen window feature selector (PWFS), minimum redundancy
maximum relevance feature selection (mRMR), and normalized MI feature selection
(NMIFS), can all adopt UFT for pre-processing of non-numerical features. Unlike
traditional FT methods, the proposed UFT is unbiased while PWFS is utilized to
its full advantage. Simulations and analyses of large-scale datasets showed
that feature subset selected by the integrated method, UFT-PWFS, outperformed
other FT-FS integrated methods in classification accuracy.
| [
{
"version": "v1",
"created": "Mon, 24 Nov 2014 10:15:17 GMT"
},
{
"version": "v2",
"created": "Sun, 29 Mar 2015 05:32:50 GMT"
}
] | 2015-03-31T00:00:00 | [
[
"Wei",
"Min",
""
],
[
"Chow",
"Tommy W. S.",
""
],
[
"Chan",
"Rosa H. M.",
""
]
] | TITLE: Mutual Information-Based Unsupervised Feature Transformation for
Heterogeneous Feature Subset Selection
ABSTRACT: Conventional mutual information (MI) based feature selection (FS) methods are
unable to handle heterogeneous feature subset selection properly because of
data format differences or estimation methods of MI between feature subset and
class label. A way to solve this problem is feature transformation (FT). In
this study, a novel unsupervised feature transformation (UFT) which can
transform non-numerical features into numerical features is developed and
tested. The UFT process is MI-based and independent of class label. MI-based FS
algorithms, such as Parzen window feature selector (PWFS), minimum redundancy
maximum relevance feature selection (mRMR), and normalized MI feature selection
(NMIFS), can all adopt UFT for pre-processing of non-numerical features. Unlike
traditional FT methods, the proposed UFT is unbiased while PWFS is utilized to
its full advantage. Simulations and analyses of large-scale datasets showed
that feature subset selected by the integrated method, UFT-PWFS, outperformed
other FT-FS integrated methods in classification accuracy.
| no_new_dataset | 0.948775 |
1503.01800 | Samira Ebrahimi Kahou | Samira Ebrahimi Kahou, Xavier Bouthillier, Pascal Lamblin, Caglar
Gulcehre, Vincent Michalski, Kishore Konda, S\'ebastien Jean, Pierre
Froumenty, Yann Dauphin, Nicolas Boulanger-Lewandowski, Raul Chandias
Ferrari, Mehdi Mirza, David Warde-Farley, Aaron Courville, Pascal Vincent,
Roland Memisevic, Christopher Pal, Yoshua Bengio | EmoNets: Multimodal deep learning approaches for emotion recognition in
video | null | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The task of the emotion recognition in the wild (EmotiW) Challenge is to
assign one of seven emotions to short video clips extracted from Hollywood
style movies. The videos depict acted-out emotions under realistic conditions
with a large degree of variation in attributes such as pose and illumination,
making it worthwhile to explore approaches which consider combinations of
features from multiple modalities for label assignment. In this paper we
present our approach to learning several specialist models using deep learning
techniques, each focusing on one modality. Among these are a convolutional
neural network, focusing on capturing visual information in detected faces, a
deep belief net focusing on the representation of the audio stream, a K-Means
based "bag-of-mouths" model, which extracts visual features around the mouth
region and a relational autoencoder, which addresses spatio-temporal aspects of
videos. We explore multiple methods for the combination of cues from these
modalities into one common classifier. This achieves a considerably greater
accuracy than predictions from our strongest single-modality classifier. Our
method was the winning submission in the 2013 EmotiW challenge and achieved a
test set accuracy of 47.67% on the 2014 dataset.
| [
{
"version": "v1",
"created": "Thu, 5 Mar 2015 22:03:26 GMT"
},
{
"version": "v2",
"created": "Mon, 30 Mar 2015 00:55:02 GMT"
}
] | 2015-03-31T00:00:00 | [
[
"Kahou",
"Samira Ebrahimi",
""
],
[
"Bouthillier",
"Xavier",
""
],
[
"Lamblin",
"Pascal",
""
],
[
"Gulcehre",
"Caglar",
""
],
[
"Michalski",
"Vincent",
""
],
[
"Konda",
"Kishore",
""
],
[
"Jean",
"Sébastien",
""
],
[
"Froumenty",
"Pierre",
""
],
[
"Dauphin",
"Yann",
""
],
[
"Boulanger-Lewandowski",
"Nicolas",
""
],
[
"Ferrari",
"Raul Chandias",
""
],
[
"Mirza",
"Mehdi",
""
],
[
"Warde-Farley",
"David",
""
],
[
"Courville",
"Aaron",
""
],
[
"Vincent",
"Pascal",
""
],
[
"Memisevic",
"Roland",
""
],
[
"Pal",
"Christopher",
""
],
[
"Bengio",
"Yoshua",
""
]
] | TITLE: EmoNets: Multimodal deep learning approaches for emotion recognition in
video
ABSTRACT: The task of the emotion recognition in the wild (EmotiW) Challenge is to
assign one of seven emotions to short video clips extracted from Hollywood
style movies. The videos depict acted-out emotions under realistic conditions
with a large degree of variation in attributes such as pose and illumination,
making it worthwhile to explore approaches which consider combinations of
features from multiple modalities for label assignment. In this paper we
present our approach to learning several specialist models using deep learning
techniques, each focusing on one modality. Among these are a convolutional
neural network, focusing on capturing visual information in detected faces, a
deep belief net focusing on the representation of the audio stream, a K-Means
based "bag-of-mouths" model, which extracts visual features around the mouth
region and a relational autoencoder, which addresses spatio-temporal aspects of
videos. We explore multiple methods for the combination of cues from these
modalities into one common classifier. This achieves a considerably greater
accuracy than predictions from our strongest single-modality classifier. Our
method was the winning submission in the 2013 EmotiW challenge and achieved a
test set accuracy of 47.67% on the 2014 dataset.
| no_new_dataset | 0.948585 |
1503.08263 | Chunhua Shen | Fayao Liu, Guosheng Lin, Chunhua Shen | CRF Learning with CNN Features for Image Segmentation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Conditional Random Rields (CRF) have been widely applied in image
segmentations. While most studies rely on hand-crafted features, we here
propose to exploit a pre-trained large convolutional neural network (CNN) to
generate deep features for CRF learning. The deep CNN is trained on the
ImageNet dataset and transferred to image segmentations here for constructing
potentials of superpixels. Then the CRF parameters are learnt using a
structured support vector machine (SSVM). To fully exploit context information
in inference, we construct spatially related co-occurrence pairwise potentials
and incorporate them into the energy function. This prefers labelling of object
pairs that frequently co-occur in a certain spatial layout and at the same time
avoids implausible labellings during the inference. Extensive experiments on
binary and multi-class segmentation benchmarks demonstrate the promise of the
proposed method. We thus provide new baselines for the segmentation performance
on the Weizmann horse, Graz-02, MSRC-21, Stanford Background and PASCAL VOC
2011 datasets.
| [
{
"version": "v1",
"created": "Sat, 28 Mar 2015 04:05:09 GMT"
}
] | 2015-03-31T00:00:00 | [
[
"Liu",
"Fayao",
""
],
[
"Lin",
"Guosheng",
""
],
[
"Shen",
"Chunhua",
""
]
] | TITLE: CRF Learning with CNN Features for Image Segmentation
ABSTRACT: Conditional Random Rields (CRF) have been widely applied in image
segmentations. While most studies rely on hand-crafted features, we here
propose to exploit a pre-trained large convolutional neural network (CNN) to
generate deep features for CRF learning. The deep CNN is trained on the
ImageNet dataset and transferred to image segmentations here for constructing
potentials of superpixels. Then the CRF parameters are learnt using a
structured support vector machine (SSVM). To fully exploit context information
in inference, we construct spatially related co-occurrence pairwise potentials
and incorporate them into the energy function. This prefers labelling of object
pairs that frequently co-occur in a certain spatial layout and at the same time
avoids implausible labellings during the inference. Extensive experiments on
binary and multi-class segmentation benchmarks demonstrate the promise of the
proposed method. We thus provide new baselines for the segmentation performance
on the Weizmann horse, Graz-02, MSRC-21, Stanford Background and PASCAL VOC
2011 datasets.
| no_new_dataset | 0.949059 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.