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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1611.07573 | Manuel Isaac Martinez Torres | Manuel Martinez, Monica Haurilet, Ziad Al-Halah, Makarand Tapaswi,
Rainer Stiefelhagen | Relaxed Earth Mover's Distances for Chain- and Tree-connected Spaces and
their use as a Loss Function in Deep Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Earth Mover's Distance (EMD) computes the optimal cost of transforming
one distribution into another, given a known transport metric between them. In
deep learning, the EMD loss allows us to embed information during training
about the output space structure like hierarchical or semantic relations. This
helps in achieving better output smoothness and generalization. However EMD is
computationally expensive.Moreover, solving EMD optimization problems usually
require complex techniques like lasso. These properties limit the applicability
of EMD-based approaches in large scale machine learning.
We address in this work the difficulties facing incorporation of EMD-based
loss in deep learning frameworks. Additionally, we provide insight and novel
solutions on how to integrate such loss function in training deep neural
networks. Specifically, we make three main contributions: (i) we provide an
in-depth analysis of the fastest state-of-the-art EMD algorithm (Sinkhorn
Distance) and discuss its limitations in deep learning scenarios. (ii) we
derive fast and numerically stable closed-form solutions for the EMD gradient
in output spaces with chain- and tree- connectivity; and (iii) we propose a
relaxed form of the EMD gradient with equivalent computational complexity but
faster convergence rate. We support our claims with experiments on real
datasets. In a restricted data setting on the ImageNet dataset, we train a
model to classify 1000 categories using 50K images, and demonstrate that our
relaxed EMD loss achieves better Top-1 accuracy than the cross entropy loss.
Overall, we show that our relaxed EMD loss criterion is a powerful asset for
deep learning in the small data regime.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 22:54:05 GMT"
}
] | 2016-11-24T00:00:00 | [
[
"Martinez",
"Manuel",
""
],
[
"Haurilet",
"Monica",
""
],
[
"Al-Halah",
"Ziad",
""
],
[
"Tapaswi",
"Makarand",
""
],
[
"Stiefelhagen",
"Rainer",
""
]
] | TITLE: Relaxed Earth Mover's Distances for Chain- and Tree-connected Spaces and
their use as a Loss Function in Deep Learning
ABSTRACT: The Earth Mover's Distance (EMD) computes the optimal cost of transforming
one distribution into another, given a known transport metric between them. In
deep learning, the EMD loss allows us to embed information during training
about the output space structure like hierarchical or semantic relations. This
helps in achieving better output smoothness and generalization. However EMD is
computationally expensive.Moreover, solving EMD optimization problems usually
require complex techniques like lasso. These properties limit the applicability
of EMD-based approaches in large scale machine learning.
We address in this work the difficulties facing incorporation of EMD-based
loss in deep learning frameworks. Additionally, we provide insight and novel
solutions on how to integrate such loss function in training deep neural
networks. Specifically, we make three main contributions: (i) we provide an
in-depth analysis of the fastest state-of-the-art EMD algorithm (Sinkhorn
Distance) and discuss its limitations in deep learning scenarios. (ii) we
derive fast and numerically stable closed-form solutions for the EMD gradient
in output spaces with chain- and tree- connectivity; and (iii) we propose a
relaxed form of the EMD gradient with equivalent computational complexity but
faster convergence rate. We support our claims with experiments on real
datasets. In a restricted data setting on the ImageNet dataset, we train a
model to classify 1000 categories using 50K images, and demonstrate that our
relaxed EMD loss achieves better Top-1 accuracy than the cross entropy loss.
Overall, we show that our relaxed EMD loss criterion is a powerful asset for
deep learning in the small data regime.
| no_new_dataset | 0.949106 |
1611.07579 | Sameer Singh | Sameer Singh and Marco Tulio Ribeiro and Carlos Guestrin | Programs as Black-Box Explanations | Presented at NIPS 2016 Workshop on Interpretable Machine Learning in
Complex Systems | null | null | null | stat.ML cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent work in model-agnostic explanations of black-box machine learning has
demonstrated that interpretability of complex models does not have to come at
the cost of accuracy or model flexibility. However, it is not clear what kind
of explanations, such as linear models, decision trees, and rule lists, are the
appropriate family to consider, and different tasks and models may benefit from
different kinds of explanations. Instead of picking a single family of
representations, in this work we propose to use "programs" as model-agnostic
explanations. We show that small programs can be expressive yet intuitive as
explanations, and generalize over a number of existing interpretable families.
We propose a prototype program induction method based on simulated annealing
that approximates the local behavior of black-box classifiers around a specific
prediction using random perturbations. Finally, we present preliminary
application on small datasets and show that the generated explanations are
intuitive and accurate for a number of classifiers.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 23:35:03 GMT"
}
] | 2016-11-24T00:00:00 | [
[
"Singh",
"Sameer",
""
],
[
"Ribeiro",
"Marco Tulio",
""
],
[
"Guestrin",
"Carlos",
""
]
] | TITLE: Programs as Black-Box Explanations
ABSTRACT: Recent work in model-agnostic explanations of black-box machine learning has
demonstrated that interpretability of complex models does not have to come at
the cost of accuracy or model flexibility. However, it is not clear what kind
of explanations, such as linear models, decision trees, and rule lists, are the
appropriate family to consider, and different tasks and models may benefit from
different kinds of explanations. Instead of picking a single family of
representations, in this work we propose to use "programs" as model-agnostic
explanations. We show that small programs can be expressive yet intuitive as
explanations, and generalize over a number of existing interpretable families.
We propose a prototype program induction method based on simulated annealing
that approximates the local behavior of black-box classifiers around a specific
prediction using random perturbations. Finally, we present preliminary
application on small datasets and show that the generated explanations are
intuitive and accurate for a number of classifiers.
| no_new_dataset | 0.945197 |
1611.07623 | EPTCS | Maaz Bin Safeer Ahmad, Alvin Cheung | Leveraging Parallel Data Processing Frameworks with Verified Lifting | In Proceedings SYNT 2016, arXiv:1611.07178 | EPTCS 229, 2016, pp. 67-83 | 10.4204/EPTCS.229.7 | null | cs.PL cs.DB cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many parallel data frameworks have been proposed in recent years that let
sequential programs access parallel processing. To capitalize on the benefits
of such frameworks, existing code must often be rewritten to the
domain-specific languages that each framework supports. This rewriting-tedious
and error-prone-also requires developers to choose the framework that best
optimizes performance given a specific workload.
This paper describes Casper, a novel compiler that automatically retargets
sequential Java code for execution on Hadoop, a parallel data processing
framework that implements the MapReduce paradigm. Given a sequential code
fragment, Casper uses verified lifting to infer a high-level summary expressed
in our program specification language that is then compiled for execution on
Hadoop. We demonstrate that Casper automatically translates Java benchmarks
into Hadoop. The translated results execute on average 3.3x faster than the
sequential implementations and scale better, as well, to larger datasets.
| [
{
"version": "v1",
"created": "Wed, 23 Nov 2016 03:16:38 GMT"
}
] | 2016-11-24T00:00:00 | [
[
"Ahmad",
"Maaz Bin Safeer",
""
],
[
"Cheung",
"Alvin",
""
]
] | TITLE: Leveraging Parallel Data Processing Frameworks with Verified Lifting
ABSTRACT: Many parallel data frameworks have been proposed in recent years that let
sequential programs access parallel processing. To capitalize on the benefits
of such frameworks, existing code must often be rewritten to the
domain-specific languages that each framework supports. This rewriting-tedious
and error-prone-also requires developers to choose the framework that best
optimizes performance given a specific workload.
This paper describes Casper, a novel compiler that automatically retargets
sequential Java code for execution on Hadoop, a parallel data processing
framework that implements the MapReduce paradigm. Given a sequential code
fragment, Casper uses verified lifting to infer a high-level summary expressed
in our program specification language that is then compiled for execution on
Hadoop. We demonstrate that Casper automatically translates Java benchmarks
into Hadoop. The translated results execute on average 3.3x faster than the
sequential implementations and scale better, as well, to larger datasets.
| no_new_dataset | 0.944995 |
1611.07675 | Ting Yao | Yingwei Pan, Ting Yao, Houqiang Li, Tao Mei | Video Captioning with Transferred Semantic Attributes | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatically generating natural language descriptions of videos plays a
fundamental challenge for computer vision community. Most recent progress in
this problem has been achieved through employing 2-D and/or 3-D Convolutional
Neural Networks (CNN) to encode video content and Recurrent Neural Networks
(RNN) to decode a sentence. In this paper, we present Long Short-Term Memory
with Transferred Semantic Attributes (LSTM-TSA)---a novel deep architecture
that incorporates the transferred semantic attributes learnt from images and
videos into the CNN plus RNN framework, by training them in an end-to-end
manner. The design of LSTM-TSA is highly inspired by the facts that 1) semantic
attributes play a significant contribution to captioning, and 2) images and
videos carry complementary semantics and thus can reinforce each other for
captioning. To boost video captioning, we propose a novel transfer unit to
model the mutually correlated attributes learnt from images and videos.
Extensive experiments are conducted on three public datasets, i.e., MSVD, M-VAD
and MPII-MD. Our proposed LSTM-TSA achieves to-date the best published
performance in sentence generation on MSVD: 52.8% and 74.0% in terms of BLEU@4
and CIDEr-D. Superior results when compared to state-of-the-art methods are
also reported on M-VAD and MPII-MD.
| [
{
"version": "v1",
"created": "Wed, 23 Nov 2016 07:59:59 GMT"
}
] | 2016-11-24T00:00:00 | [
[
"Pan",
"Yingwei",
""
],
[
"Yao",
"Ting",
""
],
[
"Li",
"Houqiang",
""
],
[
"Mei",
"Tao",
""
]
] | TITLE: Video Captioning with Transferred Semantic Attributes
ABSTRACT: Automatically generating natural language descriptions of videos plays a
fundamental challenge for computer vision community. Most recent progress in
this problem has been achieved through employing 2-D and/or 3-D Convolutional
Neural Networks (CNN) to encode video content and Recurrent Neural Networks
(RNN) to decode a sentence. In this paper, we present Long Short-Term Memory
with Transferred Semantic Attributes (LSTM-TSA)---a novel deep architecture
that incorporates the transferred semantic attributes learnt from images and
videos into the CNN plus RNN framework, by training them in an end-to-end
manner. The design of LSTM-TSA is highly inspired by the facts that 1) semantic
attributes play a significant contribution to captioning, and 2) images and
videos carry complementary semantics and thus can reinforce each other for
captioning. To boost video captioning, we propose a novel transfer unit to
model the mutually correlated attributes learnt from images and videos.
Extensive experiments are conducted on three public datasets, i.e., MSVD, M-VAD
and MPII-MD. Our proposed LSTM-TSA achieves to-date the best published
performance in sentence generation on MSVD: 52.8% and 74.0% in terms of BLEU@4
and CIDEr-D. Superior results when compared to state-of-the-art methods are
also reported on M-VAD and MPII-MD.
| no_new_dataset | 0.949902 |
1611.07743 | Gil Keren | Gil Keren, Sivan Sabato, Bj\"orn Schuller | Tunable Sensitivity to Large Errors in Neural Network Training | The paper is accepted to the AAAI 2017 conference | null | null | null | stat.ML cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When humans learn a new concept, they might ignore examples that they cannot
make sense of at first, and only later focus on such examples, when they are
more useful for learning. We propose incorporating this idea of tunable
sensitivity for hard examples in neural network learning, using a new
generalization of the cross-entropy gradient step, which can be used in place
of the gradient in any gradient-based training method. The generalized gradient
is parameterized by a value that controls the sensitivity of the training
process to harder training examples. We tested our method on several benchmark
datasets. We propose, and corroborate in our experiments, that the optimal
level of sensitivity to hard example is positively correlated with the depth of
the network. Moreover, the test prediction error obtained by our method is
generally lower than that of the vanilla cross-entropy gradient learner. We
therefore conclude that tunable sensitivity can be helpful for neural network
learning.
| [
{
"version": "v1",
"created": "Wed, 23 Nov 2016 11:14:01 GMT"
}
] | 2016-11-24T00:00:00 | [
[
"Keren",
"Gil",
""
],
[
"Sabato",
"Sivan",
""
],
[
"Schuller",
"Björn",
""
]
] | TITLE: Tunable Sensitivity to Large Errors in Neural Network Training
ABSTRACT: When humans learn a new concept, they might ignore examples that they cannot
make sense of at first, and only later focus on such examples, when they are
more useful for learning. We propose incorporating this idea of tunable
sensitivity for hard examples in neural network learning, using a new
generalization of the cross-entropy gradient step, which can be used in place
of the gradient in any gradient-based training method. The generalized gradient
is parameterized by a value that controls the sensitivity of the training
process to harder training examples. We tested our method on several benchmark
datasets. We propose, and corroborate in our experiments, that the optimal
level of sensitivity to hard example is positively correlated with the depth of
the network. Moreover, the test prediction error obtained by our method is
generally lower than that of the vanilla cross-entropy gradient learner. We
therefore conclude that tunable sensitivity can be helpful for neural network
learning.
| no_new_dataset | 0.951684 |
1611.07804 | Nikita Astrakhantsev | N. Astrakhantsev | ATR4S: Toolkit with State-of-the-art Automatic Terms Recognition Methods
in Scala | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatically recognized terminology is widely used for various
domain-specific texts processing tasks, such as machine translation,
information retrieval or sentiment analysis. However, there is still no
agreement on which methods are best suited for particular settings and,
moreover, there is no reliable comparison of already developed methods. We
believe that one of the main reasons is the lack of state-of-the-art methods
implementations, which are usually non-trivial to recreate. In order to address
these issues, we present ATR4S, an open-source software written in Scala that
comprises more than 15 methods for automatic terminology recognition (ATR) and
implements the whole pipeline from text document preprocessing, to term
candidates collection, term candidates scoring, and finally, term candidates
ranking. It is highly scalable, modular and configurable tool with support of
automatic caching. We also compare 10 state-of-the-art methods on 7 open
datasets by average precision and processing time. Experimental comparison
reveals that no single method demonstrates best average precision for all
datasets and that other available tools for ATR do not contain the best
methods.
| [
{
"version": "v1",
"created": "Wed, 23 Nov 2016 14:14:52 GMT"
}
] | 2016-11-24T00:00:00 | [
[
"Astrakhantsev",
"N.",
""
]
] | TITLE: ATR4S: Toolkit with State-of-the-art Automatic Terms Recognition Methods
in Scala
ABSTRACT: Automatically recognized terminology is widely used for various
domain-specific texts processing tasks, such as machine translation,
information retrieval or sentiment analysis. However, there is still no
agreement on which methods are best suited for particular settings and,
moreover, there is no reliable comparison of already developed methods. We
believe that one of the main reasons is the lack of state-of-the-art methods
implementations, which are usually non-trivial to recreate. In order to address
these issues, we present ATR4S, an open-source software written in Scala that
comprises more than 15 methods for automatic terminology recognition (ATR) and
implements the whole pipeline from text document preprocessing, to term
candidates collection, term candidates scoring, and finally, term candidates
ranking. It is highly scalable, modular and configurable tool with support of
automatic caching. We also compare 10 state-of-the-art methods on 7 open
datasets by average precision and processing time. Experimental comparison
reveals that no single method demonstrates best average precision for all
datasets and that other available tools for ATR do not contain the best
methods.
| no_new_dataset | 0.933915 |
1510.00936 | Ali Zarezade | Ali Zarezade, Ali Khodadadi, Mehrdad Farajtabar, Hamid R. Rabiee,
Hongyuan Zha | Correlated Cascades: Compete or Cooperate | null | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In real world social networks, there are multiple cascades which are rarely
independent. They usually compete or cooperate with each other. Motivated by
the reinforcement theory in sociology we leverage the fact that adoption of a
user to any behavior is modeled by the aggregation of behaviors of its
neighbors. We use a multidimensional marked Hawkes process to model users
product adoption and consequently spread of cascades in social networks. The
resulting inference problem is proved to be convex and is solved in parallel by
using the barrier method. The advantage of the proposed model is twofold; it
models correlated cascades and also learns the latent diffusion network.
Experimental results on synthetic and two real datasets gathered from Twitter,
URL shortening and music streaming services, illustrate the superior
performance of the proposed model over the alternatives.
| [
{
"version": "v1",
"created": "Sun, 4 Oct 2015 12:54:29 GMT"
},
{
"version": "v2",
"created": "Wed, 18 Nov 2015 16:19:52 GMT"
},
{
"version": "v3",
"created": "Tue, 22 Nov 2016 06:33:56 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Zarezade",
"Ali",
""
],
[
"Khodadadi",
"Ali",
""
],
[
"Farajtabar",
"Mehrdad",
""
],
[
"Rabiee",
"Hamid R.",
""
],
[
"Zha",
"Hongyuan",
""
]
] | TITLE: Correlated Cascades: Compete or Cooperate
ABSTRACT: In real world social networks, there are multiple cascades which are rarely
independent. They usually compete or cooperate with each other. Motivated by
the reinforcement theory in sociology we leverage the fact that adoption of a
user to any behavior is modeled by the aggregation of behaviors of its
neighbors. We use a multidimensional marked Hawkes process to model users
product adoption and consequently spread of cascades in social networks. The
resulting inference problem is proved to be convex and is solved in parallel by
using the barrier method. The advantage of the proposed model is twofold; it
models correlated cascades and also learns the latent diffusion network.
Experimental results on synthetic and two real datasets gathered from Twitter,
URL shortening and music streaming services, illustrate the superior
performance of the proposed model over the alternatives.
| no_new_dataset | 0.947478 |
1609.02372 | Thomas Gastine | T. Gastine, J. Wicht, J. Aubert | Scaling regimes in spherical shell rotating convection | 42 pages, 20 figures, 3 tables, accepted for publication in JFM | null | 10.1017/jfm.2016.659 | null | physics.flu-dyn astro-ph.EP astro-ph.SR physics.geo-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rayleigh-B\'enard convection in rotating spherical shells can be considered
as a simplified analogue of many astrophysical and geophysical fluid flows.
Here, we use three-dimensional direct numerical simulations to study this
physical process. We construct a dataset of more than 200 numerical models that
cover a broad parameter range with Ekman numbers spanning $3\times 10^{-7} \leq
E \leq 10^{-1}$, Rayleigh numbers within the range $10^3 < Ra < 2\times
10^{10}$ and a Prandtl number unity. We investigate the scaling behaviours of
both local (length scales, boundary layers) and global (Nusselt and Reynolds
numbers) properties across various physical regimes from onset of rotating
convection to weakly-rotating convection. Close to critical, the convective
flow is dominated by a triple force balance between viscosity, Coriolis force
and buoyancy. For larger supercriticalities, a subset of our numerical data
approaches the asymptotic diffusivity-free scaling of rotating convection
$Nu\sim Ra^{3/2}E^{2}$ in a narrow fraction of the parameter space delimited by
$6\,Ra_c \leq Ra \leq 0.4\,E^{-8/5}$. Using a decomposition of the viscous
dissipation rate into bulk and boundary layer contributions, we establish a
theoretical scaling of the flow velocity that accurately describes the
numerical data. In rapidly-rotating turbulent convection, the fluid bulk is
controlled by a triple force balance between Coriolis, inertia and buoyancy,
while the remaining fraction of the dissipation can be attributed to the
viscous friction in the Ekman layers. Beyond $Ra \simeq E^{-8/5}$, the
rotational constraint on the convective flow is gradually lost and the flow
properties vary to match the regime changes between rotation-dominated and
non-rotating convection. The quantity $Ra E^{12/7}$ provides an accurate
transition parameter to separate rotating and non-rotating convection.
| [
{
"version": "v1",
"created": "Thu, 8 Sep 2016 10:46:27 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Oct 2016 12:40:32 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Gastine",
"T.",
""
],
[
"Wicht",
"J.",
""
],
[
"Aubert",
"J.",
""
]
] | TITLE: Scaling regimes in spherical shell rotating convection
ABSTRACT: Rayleigh-B\'enard convection in rotating spherical shells can be considered
as a simplified analogue of many astrophysical and geophysical fluid flows.
Here, we use three-dimensional direct numerical simulations to study this
physical process. We construct a dataset of more than 200 numerical models that
cover a broad parameter range with Ekman numbers spanning $3\times 10^{-7} \leq
E \leq 10^{-1}$, Rayleigh numbers within the range $10^3 < Ra < 2\times
10^{10}$ and a Prandtl number unity. We investigate the scaling behaviours of
both local (length scales, boundary layers) and global (Nusselt and Reynolds
numbers) properties across various physical regimes from onset of rotating
convection to weakly-rotating convection. Close to critical, the convective
flow is dominated by a triple force balance between viscosity, Coriolis force
and buoyancy. For larger supercriticalities, a subset of our numerical data
approaches the asymptotic diffusivity-free scaling of rotating convection
$Nu\sim Ra^{3/2}E^{2}$ in a narrow fraction of the parameter space delimited by
$6\,Ra_c \leq Ra \leq 0.4\,E^{-8/5}$. Using a decomposition of the viscous
dissipation rate into bulk and boundary layer contributions, we establish a
theoretical scaling of the flow velocity that accurately describes the
numerical data. In rapidly-rotating turbulent convection, the fluid bulk is
controlled by a triple force balance between Coriolis, inertia and buoyancy,
while the remaining fraction of the dissipation can be attributed to the
viscous friction in the Ekman layers. Beyond $Ra \simeq E^{-8/5}$, the
rotational constraint on the convective flow is gradually lost and the flow
properties vary to match the regime changes between rotation-dominated and
non-rotating convection. The quantity $Ra E^{12/7}$ provides an accurate
transition parameter to separate rotating and non-rotating convection.
| no_new_dataset | 0.942718 |
1611.04326 | Aditya Joshi | Aditya Joshi, Prayas Jain, Pushpak Bhattacharyya, Mark Carman | `Who would have thought of that!': A Hierarchical Topic Model for
Extraction of Sarcasm-prevalent Topics and Sarcasm Detection | This version of the paper contains corrected changes, after the
camera -ready submission. These changes were observed based on an issue in
the output returned by SVM Perf. This paper will be presented at ExPROM
workshop at COLING 2016 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Topic Models have been reported to be beneficial for aspect-based sentiment
analysis. This paper reports a simple topic model for sarcasm detection, a
first, to the best of our knowledge. Designed on the basis of the intuition
that sarcastic tweets are likely to have a mixture of words of both sentiments
as against tweets with literal sentiment (either positive or negative), our
hierarchical topic model discovers sarcasm-prevalent topics and topic-level
sentiment. Using a dataset of tweets labeled using hashtags, the model
estimates topic-level, and sentiment-level distributions. Our evaluation shows
that topics such as `work', `gun laws', `weather' are sarcasm-prevalent topics.
Our model is also able to discover the mixture of sentiment-bearing words that
exist in a text of a given sentiment-related label. Finally, we apply our model
to predict sarcasm in tweets. We outperform two prior work based on statistical
classifiers with specific features, by around 25\%.
| [
{
"version": "v1",
"created": "Mon, 14 Nov 2016 10:40:44 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Nov 2016 10:55:56 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Joshi",
"Aditya",
""
],
[
"Jain",
"Prayas",
""
],
[
"Bhattacharyya",
"Pushpak",
""
],
[
"Carman",
"Mark",
""
]
] | TITLE: `Who would have thought of that!': A Hierarchical Topic Model for
Extraction of Sarcasm-prevalent Topics and Sarcasm Detection
ABSTRACT: Topic Models have been reported to be beneficial for aspect-based sentiment
analysis. This paper reports a simple topic model for sarcasm detection, a
first, to the best of our knowledge. Designed on the basis of the intuition
that sarcastic tweets are likely to have a mixture of words of both sentiments
as against tweets with literal sentiment (either positive or negative), our
hierarchical topic model discovers sarcasm-prevalent topics and topic-level
sentiment. Using a dataset of tweets labeled using hashtags, the model
estimates topic-level, and sentiment-level distributions. Our evaluation shows
that topics such as `work', `gun laws', `weather' are sarcasm-prevalent topics.
Our model is also able to discover the mixture of sentiment-bearing words that
exist in a text of a given sentiment-related label. Finally, we apply our model
to predict sarcasm in tweets. We outperform two prior work based on statistical
classifiers with specific features, by around 25\%.
| no_new_dataset | 0.948298 |
1611.05244 | Mengyue Geng | Mengyue Geng and Yaowei Wang and Tao Xiang and Yonghong Tian | Deep Transfer Learning for Person Re-identification | 12 pages, 2 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Person re-identification (Re-ID) poses a unique challenge to deep learning:
how to learn a deep model with millions of parameters on a small training set
of few or no labels. In this paper, a number of deep transfer learning models
are proposed to address the data sparsity problem. First, a deep network
architecture is designed which differs from existing deep Re-ID models in that
(a) it is more suitable for transferring representations learned from large
image classification datasets, and (b) classification loss and verification
loss are combined, each of which adopts a different dropout strategy. Second, a
two-stepped fine-tuning strategy is developed to transfer knowledge from
auxiliary datasets. Third, given an unlabelled Re-ID dataset, a novel
unsupervised deep transfer learning model is developed based on co-training.
The proposed models outperform the state-of-the-art deep Re-ID models by large
margins: we achieve Rank-1 accuracy of 85.4\%, 83.7\% and 56.3\% on CUHK03,
Market1501, and VIPeR respectively, whilst on VIPeR, our unsupervised model
(45.1\%) beats most supervised models.
| [
{
"version": "v1",
"created": "Wed, 16 Nov 2016 12:14:09 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Nov 2016 12:16:51 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Geng",
"Mengyue",
""
],
[
"Wang",
"Yaowei",
""
],
[
"Xiang",
"Tao",
""
],
[
"Tian",
"Yonghong",
""
]
] | TITLE: Deep Transfer Learning for Person Re-identification
ABSTRACT: Person re-identification (Re-ID) poses a unique challenge to deep learning:
how to learn a deep model with millions of parameters on a small training set
of few or no labels. In this paper, a number of deep transfer learning models
are proposed to address the data sparsity problem. First, a deep network
architecture is designed which differs from existing deep Re-ID models in that
(a) it is more suitable for transferring representations learned from large
image classification datasets, and (b) classification loss and verification
loss are combined, each of which adopts a different dropout strategy. Second, a
two-stepped fine-tuning strategy is developed to transfer knowledge from
auxiliary datasets. Third, given an unlabelled Re-ID dataset, a novel
unsupervised deep transfer learning model is developed based on co-training.
The proposed models outperform the state-of-the-art deep Re-ID models by large
margins: we achieve Rank-1 accuracy of 85.4\%, 83.7\% and 56.3\% on CUHK03,
Market1501, and VIPeR respectively, whilst on VIPeR, our unsupervised model
(45.1\%) beats most supervised models.
| no_new_dataset | 0.952042 |
1611.07054 | Sebastian P\"olsterl | Sebastian P\"olsterl, Nassir Navab, Amin Katouzian | An Efficient Training Algorithm for Kernel Survival Support Vector
Machines | ECML PKDD MLLS 2016: 3rd Workshop on Machine Learning in Life
Sciences | null | null | null | cs.LG cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Survival analysis is a fundamental tool in medical research to identify
predictors of adverse events and develop systems for clinical decision support.
In order to leverage large amounts of patient data, efficient optimisation
routines are paramount. We propose an efficient training algorithm for the
kernel survival support vector machine (SSVM). We directly optimise the primal
objective function and employ truncated Newton optimisation and order statistic
trees to significantly lower computational costs compared to previous training
algorithms, which require $O(n^4)$ space and $O(p n^6)$ time for datasets with
$n$ samples and $p$ features. Our results demonstrate that our proposed
optimisation scheme allows analysing data of a much larger scale with no loss
in prediction performance. Experiments on synthetic and 5 real-world datasets
show that our technique outperforms existing kernel SSVM formulations if the
amount of right censoring is high ($\geq85\%$), and performs comparably
otherwise.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 21:09:33 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Pölsterl",
"Sebastian",
""
],
[
"Navab",
"Nassir",
""
],
[
"Katouzian",
"Amin",
""
]
] | TITLE: An Efficient Training Algorithm for Kernel Survival Support Vector
Machines
ABSTRACT: Survival analysis is a fundamental tool in medical research to identify
predictors of adverse events and develop systems for clinical decision support.
In order to leverage large amounts of patient data, efficient optimisation
routines are paramount. We propose an efficient training algorithm for the
kernel survival support vector machine (SSVM). We directly optimise the primal
objective function and employ truncated Newton optimisation and order statistic
trees to significantly lower computational costs compared to previous training
algorithms, which require $O(n^4)$ space and $O(p n^6)$ time for datasets with
$n$ samples and $p$ features. Our results demonstrate that our proposed
optimisation scheme allows analysing data of a much larger scale with no loss
in prediction performance. Experiments on synthetic and 5 real-world datasets
show that our technique outperforms existing kernel SSVM formulations if the
amount of right censoring is high ($\geq85\%$), and performs comparably
otherwise.
| no_new_dataset | 0.949012 |
1611.07119 | Chongxuan Li | Chongxuan Li and Jun Zhu and Bo Zhang | Max-Margin Deep Generative Models for (Semi-)Supervised Learning | arXiv admin note: substantial text overlap with arXiv:1504.06787 | null | null | null | cs.CV cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep generative models (DGMs) are effective on learning multilayered
representations of complex data and performing inference of input data by
exploring the generative ability. However, it is relatively insufficient to
empower the discriminative ability of DGMs on making accurate predictions. This
paper presents max-margin deep generative models (mmDGMs) and a
class-conditional variant (mmDCGMs), which explore the strongly discriminative
principle of max-margin learning to improve the predictive performance of DGMs
in both supervised and semi-supervised learning, while retaining the generative
capability. In semi-supervised learning, we use the predictions of a max-margin
classifier as the missing labels instead of performing full posterior inference
for efficiency; we also introduce additional max-margin and label-balance
regularization terms of unlabeled data for effectiveness. We develop an
efficient doubly stochastic subgradient algorithm for the piecewise linear
objectives in different settings. Empirical results on various datasets
demonstrate that: (1) max-margin learning can significantly improve the
prediction performance of DGMs and meanwhile retain the generative ability; (2)
in supervised learning, mmDGMs are competitive to the best fully discriminative
networks when employing convolutional neural networks as the generative and
recognition models; and (3) in semi-supervised learning, mmDCGMs can perform
efficient inference and achieve state-of-the-art classification results on
several benchmarks.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 01:36:29 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Li",
"Chongxuan",
""
],
[
"Zhu",
"Jun",
""
],
[
"Zhang",
"Bo",
""
]
] | TITLE: Max-Margin Deep Generative Models for (Semi-)Supervised Learning
ABSTRACT: Deep generative models (DGMs) are effective on learning multilayered
representations of complex data and performing inference of input data by
exploring the generative ability. However, it is relatively insufficient to
empower the discriminative ability of DGMs on making accurate predictions. This
paper presents max-margin deep generative models (mmDGMs) and a
class-conditional variant (mmDCGMs), which explore the strongly discriminative
principle of max-margin learning to improve the predictive performance of DGMs
in both supervised and semi-supervised learning, while retaining the generative
capability. In semi-supervised learning, we use the predictions of a max-margin
classifier as the missing labels instead of performing full posterior inference
for efficiency; we also introduce additional max-margin and label-balance
regularization terms of unlabeled data for effectiveness. We develop an
efficient doubly stochastic subgradient algorithm for the piecewise linear
objectives in different settings. Empirical results on various datasets
demonstrate that: (1) max-margin learning can significantly improve the
prediction performance of DGMs and meanwhile retain the generative ability; (2)
in supervised learning, mmDGMs are competitive to the best fully discriminative
networks when employing convolutional neural networks as the generative and
recognition models; and (3) in semi-supervised learning, mmDCGMs can perform
efficient inference and achieve state-of-the-art classification results on
several benchmarks.
| no_new_dataset | 0.948822 |
1611.07212 | Albert Haque | Albert Haque, Alexandre Alahi, Li Fei-Fei | Recurrent Attention Models for Depth-Based Person Identification | Computer Vision and Pattern Recognition (CVPR) 2016 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present an attention-based model that reasons on human body shape and
motion dynamics to identify individuals in the absence of RGB information,
hence in the dark. Our approach leverages unique 4D spatio-temporal signatures
to address the identification problem across days. Formulated as a
reinforcement learning task, our model is based on a combination of
convolutional and recurrent neural networks with the goal of identifying small,
discriminative regions indicative of human identity. We demonstrate that our
model produces state-of-the-art results on several published datasets given
only depth images. We further study the robustness of our model towards
viewpoint, appearance, and volumetric changes. Finally, we share insights
gleaned from interpretable 2D, 3D, and 4D visualizations of our model's
spatio-temporal attention.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 09:27:30 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Haque",
"Albert",
""
],
[
"Alahi",
"Alexandre",
""
],
[
"Fei-Fei",
"Li",
""
]
] | TITLE: Recurrent Attention Models for Depth-Based Person Identification
ABSTRACT: We present an attention-based model that reasons on human body shape and
motion dynamics to identify individuals in the absence of RGB information,
hence in the dark. Our approach leverages unique 4D spatio-temporal signatures
to address the identification problem across days. Formulated as a
reinforcement learning task, our model is based on a combination of
convolutional and recurrent neural networks with the goal of identifying small,
discriminative regions indicative of human identity. We demonstrate that our
model produces state-of-the-art results on several published datasets given
only depth images. We further study the robustness of our model towards
viewpoint, appearance, and volumetric changes. Finally, we share insights
gleaned from interpretable 2D, 3D, and 4D visualizations of our model's
spatio-temporal attention.
| no_new_dataset | 0.948346 |
1611.07308 | Thomas Kipf | Thomas N. Kipf, Max Welling | Variational Graph Auto-Encoders | Bayesian Deep Learning Workshop (NIPS 2016) | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce the variational graph auto-encoder (VGAE), a framework for
unsupervised learning on graph-structured data based on the variational
auto-encoder (VAE). This model makes use of latent variables and is capable of
learning interpretable latent representations for undirected graphs. We
demonstrate this model using a graph convolutional network (GCN) encoder and a
simple inner product decoder. Our model achieves competitive results on a link
prediction task in citation networks. In contrast to most existing models for
unsupervised learning on graph-structured data and link prediction, our model
can naturally incorporate node features, which significantly improves
predictive performance on a number of benchmark datasets.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 11:37:17 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Kipf",
"Thomas N.",
""
],
[
"Welling",
"Max",
""
]
] | TITLE: Variational Graph Auto-Encoders
ABSTRACT: We introduce the variational graph auto-encoder (VGAE), a framework for
unsupervised learning on graph-structured data based on the variational
auto-encoder (VAE). This model makes use of latent variables and is capable of
learning interpretable latent representations for undirected graphs. We
demonstrate this model using a graph convolutional network (GCN) encoder and a
simple inner product decoder. Our model achieves competitive results on a link
prediction task in citation networks. In contrast to most existing models for
unsupervised learning on graph-structured data and link prediction, our model
can naturally incorporate node features, which significantly improves
predictive performance on a number of benchmark datasets.
| no_new_dataset | 0.94699 |
1611.07385 | Xiao Yang | Xiao Yang, Dafang He, Wenyi Huang, Zihan Zhou, Alex Ororbia, Dan
Kifer, C. Lee Giles | Smart Library: Identifying Books in a Library using Richly Supervised
Deep Scene Text Reading | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Physical library collections are valuable and long standing resources for
knowledge and learning. However, managing books in a large bookshelf and
finding books on it often leads to tedious manual work, especially for large
book collections where books might be missing or misplaced. Recently, deep
neural models, such as Convolutional Neural Networks (CNN) and Recurrent Neural
Networks (RNN) have achieved great success for scene text detection and
recognition. Motivated by these recent successes, we aim to investigate their
viability in facilitating book management, a task that introduces further
challenges including large amounts of cluttered scene text, distortion, and
varied lighting conditions. In this paper, we present a library inventory
building and retrieval system based on scene text reading methods. We
specifically design our scene text recognition model using rich supervision to
accelerate training and achieve state-of-the-art performance on several
benchmark datasets. Our proposed system has the potential to greatly reduce the
amount of human labor required in managing book inventories as well as the
space needed to store book information.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 16:12:03 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Yang",
"Xiao",
""
],
[
"He",
"Dafang",
""
],
[
"Huang",
"Wenyi",
""
],
[
"Zhou",
"Zihan",
""
],
[
"Ororbia",
"Alex",
""
],
[
"Kifer",
"Dan",
""
],
[
"Giles",
"C. Lee",
""
]
] | TITLE: Smart Library: Identifying Books in a Library using Richly Supervised
Deep Scene Text Reading
ABSTRACT: Physical library collections are valuable and long standing resources for
knowledge and learning. However, managing books in a large bookshelf and
finding books on it often leads to tedious manual work, especially for large
book collections where books might be missing or misplaced. Recently, deep
neural models, such as Convolutional Neural Networks (CNN) and Recurrent Neural
Networks (RNN) have achieved great success for scene text detection and
recognition. Motivated by these recent successes, we aim to investigate their
viability in facilitating book management, a task that introduces further
challenges including large amounts of cluttered scene text, distortion, and
varied lighting conditions. In this paper, we present a library inventory
building and retrieval system based on scene text reading methods. We
specifically design our scene text recognition model using rich supervision to
accelerate training and achieve state-of-the-art performance on several
benchmark datasets. Our proposed system has the potential to greatly reduce the
amount of human labor required in managing book inventories as well as the
space needed to store book information.
| no_new_dataset | 0.948202 |
1611.07438 | Lu Zhang | Lu Zhang (1), Yongkai Wu (1), Xintao Wu (1) ((1) University of
Arkansas) | Achieving non-discrimination in data release | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Discrimination discovery and prevention/removal are increasingly important
tasks in data mining. Discrimination discovery aims to unveil discriminatory
practices on the protected attribute (e.g., gender) by analyzing the dataset of
historical decision records, and discrimination prevention aims to remove
discrimination by modifying the biased data before conducting predictive
analysis. In this paper, we show that the key to discrimination discovery and
prevention is to find the meaningful partitions that can be used to provide
quantitative evidences for the judgment of discrimination. With the support of
the causal graph, we present a graphical condition for identifying a meaningful
partition. Based on that, we develop a simple criterion for the claim of
non-discrimination, and propose discrimination removal algorithms which
accurately remove discrimination while retaining good data utility. Experiments
using real datasets show the effectiveness of our approaches.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 17:55:28 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Zhang",
"Lu",
""
],
[
"Wu",
"Yongkai",
""
],
[
"Wu",
"Xintao",
""
]
] | TITLE: Achieving non-discrimination in data release
ABSTRACT: Discrimination discovery and prevention/removal are increasingly important
tasks in data mining. Discrimination discovery aims to unveil discriminatory
practices on the protected attribute (e.g., gender) by analyzing the dataset of
historical decision records, and discrimination prevention aims to remove
discrimination by modifying the biased data before conducting predictive
analysis. In this paper, we show that the key to discrimination discovery and
prevention is to find the meaningful partitions that can be used to provide
quantitative evidences for the judgment of discrimination. With the support of
the causal graph, we present a graphical condition for identifying a meaningful
partition. Based on that, we develop a simple criterion for the claim of
non-discrimination, and propose discrimination removal algorithms which
accurately remove discrimination while retaining good data utility. Experiments
using real datasets show the effectiveness of our approaches.
| no_new_dataset | 0.950869 |
1611.07509 | Lu Zhang | Lu Zhang (1), Yongkai Wu (1), Xintao Wu (1) ((1) University of
Arkansas) | A causal framework for discovering and removing direct and indirect
discrimination | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Anti-discrimination is an increasingly important task in data science. In
this paper, we investigate the problem of discovering both direct and indirect
discrimination from the historical data, and removing the discriminatory
effects before the data is used for predictive analysis (e.g., building
classifiers). We make use of the causal network to capture the causal structure
of the data. Then we model direct and indirect discrimination as the
path-specific effects, which explicitly distinguish the two types of
discrimination as the causal effects transmitted along different paths in the
network. Based on that, we propose an effective algorithm for discovering
direct and indirect discrimination, as well as an algorithm for precisely
removing both types of discrimination while retaining good data utility.
Different from previous works, our approaches can ensure that the predictive
models built from the modified data will not incur discrimination in decision
making. Experiments using real datasets show the effectiveness of our
approaches.
| [
{
"version": "v1",
"created": "Tue, 22 Nov 2016 20:50:47 GMT"
}
] | 2016-11-23T00:00:00 | [
[
"Zhang",
"Lu",
""
],
[
"Wu",
"Yongkai",
""
],
[
"Wu",
"Xintao",
""
]
] | TITLE: A causal framework for discovering and removing direct and indirect
discrimination
ABSTRACT: Anti-discrimination is an increasingly important task in data science. In
this paper, we investigate the problem of discovering both direct and indirect
discrimination from the historical data, and removing the discriminatory
effects before the data is used for predictive analysis (e.g., building
classifiers). We make use of the causal network to capture the causal structure
of the data. Then we model direct and indirect discrimination as the
path-specific effects, which explicitly distinguish the two types of
discrimination as the causal effects transmitted along different paths in the
network. Based on that, we propose an effective algorithm for discovering
direct and indirect discrimination, as well as an algorithm for precisely
removing both types of discrimination while retaining good data utility.
Different from previous works, our approaches can ensure that the predictive
models built from the modified data will not incur discrimination in decision
making. Experiments using real datasets show the effectiveness of our
approaches.
| no_new_dataset | 0.950411 |
1411.5417 | Abhradeep Guha Thakurta | Kunal Talwar, Abhradeep Thakurta, Li Zhang | Private Empirical Risk Minimization Beyond the Worst Case: The Effect of
the Constraint Set Geometry | null | null | null | null | cs.LG cs.CR stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Empirical Risk Minimization (ERM) is a standard technique in machine
learning, where a model is selected by minimizing a loss function over
constraint set. When the training dataset consists of private information, it
is natural to use a differentially private ERM algorithm, and this problem has
been the subject of a long line of work started with Chaudhuri and Monteleoni
2008. A private ERM algorithm outputs an approximate minimizer of the loss
function and its error can be measured as the difference from the optimal value
of the loss function. When the constraint set is arbitrary, the required error
bounds are fairly well understood \cite{BassilyST14}. In this work, we show
that the geometric properties of the constraint set can be used to derive
significantly better results. Specifically, we show that a differentially
private version of Mirror Descent leads to error bounds of the form
$\tilde{O}(G_{\mathcal{C}}/n)$ for a lipschitz loss function, improving on the
$\tilde{O}(\sqrt{p}/n)$ bounds in Bassily, Smith and Thakurta 2014. Here $p$ is
the dimensionality of the problem, $n$ is the number of data points in the
training set, and $G_{\mathcal{C}}$ denotes the Gaussian width of the
constraint set that we optimize over. We show similar improvements for strongly
convex functions, and for smooth functions. In addition, we show that when the
loss function is Lipschitz with respect to the $\ell_1$ norm and $\mathcal{C}$
is $\ell_1$-bounded, a differentially private version of the Frank-Wolfe
algorithm gives error bounds of the form $\tilde{O}(n^{-2/3})$. This captures
the important and common case of sparse linear regression (LASSO), when the
data $x_i$ satisfies $|x_i|_{\infty} \leq 1$ and we optimize over the $\ell_1$
ball. We show new lower bounds for this setting, that together with known
bounds, imply that all our upper bounds are tight.
| [
{
"version": "v1",
"created": "Thu, 20 Nov 2014 01:33:53 GMT"
},
{
"version": "v2",
"created": "Thu, 5 Mar 2015 02:38:52 GMT"
},
{
"version": "v3",
"created": "Sun, 20 Nov 2016 22:40:46 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Talwar",
"Kunal",
""
],
[
"Thakurta",
"Abhradeep",
""
],
[
"Zhang",
"Li",
""
]
] | TITLE: Private Empirical Risk Minimization Beyond the Worst Case: The Effect of
the Constraint Set Geometry
ABSTRACT: Empirical Risk Minimization (ERM) is a standard technique in machine
learning, where a model is selected by minimizing a loss function over
constraint set. When the training dataset consists of private information, it
is natural to use a differentially private ERM algorithm, and this problem has
been the subject of a long line of work started with Chaudhuri and Monteleoni
2008. A private ERM algorithm outputs an approximate minimizer of the loss
function and its error can be measured as the difference from the optimal value
of the loss function. When the constraint set is arbitrary, the required error
bounds are fairly well understood \cite{BassilyST14}. In this work, we show
that the geometric properties of the constraint set can be used to derive
significantly better results. Specifically, we show that a differentially
private version of Mirror Descent leads to error bounds of the form
$\tilde{O}(G_{\mathcal{C}}/n)$ for a lipschitz loss function, improving on the
$\tilde{O}(\sqrt{p}/n)$ bounds in Bassily, Smith and Thakurta 2014. Here $p$ is
the dimensionality of the problem, $n$ is the number of data points in the
training set, and $G_{\mathcal{C}}$ denotes the Gaussian width of the
constraint set that we optimize over. We show similar improvements for strongly
convex functions, and for smooth functions. In addition, we show that when the
loss function is Lipschitz with respect to the $\ell_1$ norm and $\mathcal{C}$
is $\ell_1$-bounded, a differentially private version of the Frank-Wolfe
algorithm gives error bounds of the form $\tilde{O}(n^{-2/3})$. This captures
the important and common case of sparse linear regression (LASSO), when the
data $x_i$ satisfies $|x_i|_{\infty} \leq 1$ and we optimize over the $\ell_1$
ball. We show new lower bounds for this setting, that together with known
bounds, imply that all our upper bounds are tight.
| no_new_dataset | 0.949902 |
1509.01520 | Radu Horaud P | Sileye Ba, Xavier Alameda-Pineda, Alessio Xompero and Radu Horaud | An On-line Variational Bayesian Model for Multi-Person Tracking from
Cluttered Scenes | 21 pages, 9 figures, 4 tables | Computer Vision and Image Understanding, volume 153, December
2016, pages 64-76 | 10.1016/j.cviu.2016.07.006 | null | cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Object tracking is an ubiquitous problem that appears in many applications
such as remote sensing, audio processing, computer vision, human-machine
interfaces, human-robot interaction, etc. Although thoroughly investigated in
computer vision, tracking a time-varying number of persons remains a
challenging open problem. In this paper, we propose an on-line variational
Bayesian model for multi-person tracking from cluttered visual observations
provided by person detectors. The contributions of this paper are the
followings. First, we propose a variational Bayesian framework for tracking an
unknown and varying number of persons. Second, our model results in a
variational expectation-maximization (VEM) algorithm with closed-form
expressions for the posterior distributions of the latent variables and for the
estimation of the model parameters. Third, the proposed model exploits
observations from multiple detectors, and it is therefore multimodal by nature.
Finally, we propose to embed both object-birth and object-visibility processes
in an effort to robustly handle person appearances and disappearances over
time. Evaluated on classical multiple person tracking datasets, our method
shows competitive results with respect to state-of-the-art multiple-object
tracking models, such as the probability hypothesis density (PHD) filter among
others.
| [
{
"version": "v1",
"created": "Fri, 4 Sep 2015 16:16:42 GMT"
},
{
"version": "v2",
"created": "Thu, 31 Mar 2016 13:06:59 GMT"
},
{
"version": "v3",
"created": "Thu, 30 Jun 2016 08:50:42 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Ba",
"Sileye",
""
],
[
"Alameda-Pineda",
"Xavier",
""
],
[
"Xompero",
"Alessio",
""
],
[
"Horaud",
"Radu",
""
]
] | TITLE: An On-line Variational Bayesian Model for Multi-Person Tracking from
Cluttered Scenes
ABSTRACT: Object tracking is an ubiquitous problem that appears in many applications
such as remote sensing, audio processing, computer vision, human-machine
interfaces, human-robot interaction, etc. Although thoroughly investigated in
computer vision, tracking a time-varying number of persons remains a
challenging open problem. In this paper, we propose an on-line variational
Bayesian model for multi-person tracking from cluttered visual observations
provided by person detectors. The contributions of this paper are the
followings. First, we propose a variational Bayesian framework for tracking an
unknown and varying number of persons. Second, our model results in a
variational expectation-maximization (VEM) algorithm with closed-form
expressions for the posterior distributions of the latent variables and for the
estimation of the model parameters. Third, the proposed model exploits
observations from multiple detectors, and it is therefore multimodal by nature.
Finally, we propose to embed both object-birth and object-visibility processes
in an effort to robustly handle person appearances and disappearances over
time. Evaluated on classical multiple person tracking datasets, our method
shows competitive results with respect to state-of-the-art multiple-object
tracking models, such as the probability hypothesis density (PHD) filter among
others.
| no_new_dataset | 0.949995 |
1604.03058 | Xundong Wu | Xundong Wu, Yong Wu and Yong Zhao | Binarized Neural Networks on the ImageNet Classification Task | null | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We trained Binarized Neural Networks (BNNs) on the high resolution ImageNet
ILSVRC-2102 dataset classification task and achieved a good performance. With a
moderate size network of 13 layers, we obtained top-5 classification accuracy
rate of 84.1 % on validation set through network distillation, much better than
previous published results of 73.2% on XNOR network and 69.1% on binarized
GoogleNET. We expect networks of better performance can be obtained by
following our current strategies. We provide a detailed discussion and
preliminary analysis on strategies used in the network training.
| [
{
"version": "v1",
"created": "Mon, 11 Apr 2016 18:39:33 GMT"
},
{
"version": "v2",
"created": "Wed, 13 Apr 2016 03:22:37 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Oct 2016 15:25:06 GMT"
},
{
"version": "v4",
"created": "Tue, 8 Nov 2016 00:38:03 GMT"
},
{
"version": "v5",
"created": "Sat, 19 Nov 2016 01:37:40 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Wu",
"Xundong",
""
],
[
"Wu",
"Yong",
""
],
[
"Zhao",
"Yong",
""
]
] | TITLE: Binarized Neural Networks on the ImageNet Classification Task
ABSTRACT: We trained Binarized Neural Networks (BNNs) on the high resolution ImageNet
ILSVRC-2102 dataset classification task and achieved a good performance. With a
moderate size network of 13 layers, we obtained top-5 classification accuracy
rate of 84.1 % on validation set through network distillation, much better than
previous published results of 73.2% on XNOR network and 69.1% on binarized
GoogleNET. We expect networks of better performance can be obtained by
following our current strategies. We provide a detailed discussion and
preliminary analysis on strategies used in the network training.
| no_new_dataset | 0.958577 |
1606.03568 | Mikael K{\aa}geb\"ack | Mikael K{\aa}geb\"ack, Hans Salomonsson | Word Sense Disambiguation using a Bidirectional LSTM | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we present a clean, yet effective, model for word sense
disambiguation. Our approach leverage a bidirectional long short-term memory
network which is shared between all words. This enables the model to share
statistical strength and to scale well with vocabulary size. The model is
trained end-to-end, directly from the raw text to sense labels, and makes
effective use of word order. We evaluate our approach on two standard datasets,
using identical hyperparameter settings, which are in turn tuned on a third set
of held out data. We employ no external resources (e.g. knowledge graphs,
part-of-speech tagging, etc), language specific features, or hand crafted
rules, but still achieve statistically equivalent results to the best
state-of-the-art systems, that employ no such limitations.
| [
{
"version": "v1",
"created": "Sat, 11 Jun 2016 08:12:02 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Nov 2016 22:47:07 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Kågebäck",
"Mikael",
""
],
[
"Salomonsson",
"Hans",
""
]
] | TITLE: Word Sense Disambiguation using a Bidirectional LSTM
ABSTRACT: In this paper we present a clean, yet effective, model for word sense
disambiguation. Our approach leverage a bidirectional long short-term memory
network which is shared between all words. This enables the model to share
statistical strength and to scale well with vocabulary size. The model is
trained end-to-end, directly from the raw text to sense labels, and makes
effective use of word order. We evaluate our approach on two standard datasets,
using identical hyperparameter settings, which are in turn tuned on a third set
of held out data. We employ no external resources (e.g. knowledge graphs,
part-of-speech tagging, etc), language specific features, or hand crafted
rules, but still achieve statistically equivalent results to the best
state-of-the-art systems, that employ no such limitations.
| no_new_dataset | 0.952926 |
1606.04327 | Pawel Foremski | Pawel Foremski, David Plonka, Arthur Berger | Entropy/IP: Uncovering Structure in IPv6 Addresses | Paper presented at the ACM IMC 2016 in Santa Monica, USA
(https://dl.acm.org/citation.cfm?id=2987445). Live Demo site available at
http://www.entropy-ip.com/ | IMC '16 Proceedings of the 2016 ACM on Internet Measurement
Conference, pp. 167-181 | 10.1145/2987443.2987445 | null | cs.NI cs.AI cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we introduce Entropy/IP: a system that discovers Internet
address structure based on analyses of a subset of IPv6 addresses known to be
active, i.e., training data, gleaned by readily available passive and active
means. The system is completely automated and employs a combination of
information-theoretic and machine learning techniques to probabilistically
model IPv6 addresses. We present results showing that our system is effective
in exposing structural characteristics of portions of the IPv6 Internet address
space populated by active client, service, and router addresses.
In addition to visualizing the address structure for exploration, the system
uses its models to generate candidate target addresses for scanning. For each
of 15 evaluated datasets, we train on 1K addresses and generate 1M candidates
for scanning. We achieve some success in 14 datasets, finding up to 40% of the
generated addresses to be active. In 11 of these datasets, we find active
network identifiers (e.g., /64 prefixes or `subnets') not seen in training.
Thus, we provide the first evidence that it is practical to discover subnets
and hosts by scanning probabilistically selected areas of the IPv6 address
space not known to contain active hosts a priori.
| [
{
"version": "v1",
"created": "Tue, 14 Jun 2016 12:38:26 GMT"
},
{
"version": "v2",
"created": "Mon, 21 Nov 2016 12:43:00 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Foremski",
"Pawel",
""
],
[
"Plonka",
"David",
""
],
[
"Berger",
"Arthur",
""
]
] | TITLE: Entropy/IP: Uncovering Structure in IPv6 Addresses
ABSTRACT: In this paper, we introduce Entropy/IP: a system that discovers Internet
address structure based on analyses of a subset of IPv6 addresses known to be
active, i.e., training data, gleaned by readily available passive and active
means. The system is completely automated and employs a combination of
information-theoretic and machine learning techniques to probabilistically
model IPv6 addresses. We present results showing that our system is effective
in exposing structural characteristics of portions of the IPv6 Internet address
space populated by active client, service, and router addresses.
In addition to visualizing the address structure for exploration, the system
uses its models to generate candidate target addresses for scanning. For each
of 15 evaluated datasets, we train on 1K addresses and generate 1M candidates
for scanning. We achieve some success in 14 datasets, finding up to 40% of the
generated addresses to be active. In 11 of these datasets, we find active
network identifiers (e.g., /64 prefixes or `subnets') not seen in training.
Thus, we provide the first evidence that it is practical to discover subnets
and hosts by scanning probabilistically selected areas of the IPv6 address
space not known to contain active hosts a priori.
| no_new_dataset | 0.945751 |
1609.04909 | Shourya Roy | Shourya Roy, Himanshu S. Bhatt, Y. Narahari | An Iterative Transfer Learning Based Ensemble Technique for Automatic
Short Answer Grading | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic short answer grading (ASAG) techniques are designed to
automatically assess short answers to questions in natural language, having a
length of a few words to a few sentences. Supervised ASAG techniques have been
demonstrated to be effective but suffer from a couple of key practical
limitations. They are greatly reliant on instructor provided model answers and
need labeled training data in the form of graded student answers for every
assessment task. To overcome these, in this paper, we introduce an ASAG
technique with two novel features. We propose an iterative technique on an
ensemble of (a) a text classifier of student answers and (b) a classifier using
numeric features derived from various similarity measures with respect to model
answers. Second, we employ canonical correlation analysis based transfer
learning on a common feature representation to build the classifier ensemble
for questions having no labelled data. The proposed technique handsomely beats
all winning supervised entries on the SCIENTSBANK dataset from the Student
Response Analysis task of SemEval 2013. Additionally, we demonstrate
generalizability and benefits of the proposed technique through evaluation on
multiple ASAG datasets from different subject topics and standards.
| [
{
"version": "v1",
"created": "Fri, 16 Sep 2016 04:58:54 GMT"
},
{
"version": "v2",
"created": "Mon, 19 Sep 2016 01:28:17 GMT"
},
{
"version": "v3",
"created": "Mon, 21 Nov 2016 13:44:09 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Roy",
"Shourya",
""
],
[
"Bhatt",
"Himanshu S.",
""
],
[
"Narahari",
"Y.",
""
]
] | TITLE: An Iterative Transfer Learning Based Ensemble Technique for Automatic
Short Answer Grading
ABSTRACT: Automatic short answer grading (ASAG) techniques are designed to
automatically assess short answers to questions in natural language, having a
length of a few words to a few sentences. Supervised ASAG techniques have been
demonstrated to be effective but suffer from a couple of key practical
limitations. They are greatly reliant on instructor provided model answers and
need labeled training data in the form of graded student answers for every
assessment task. To overcome these, in this paper, we introduce an ASAG
technique with two novel features. We propose an iterative technique on an
ensemble of (a) a text classifier of student answers and (b) a classifier using
numeric features derived from various similarity measures with respect to model
answers. Second, we employ canonical correlation analysis based transfer
learning on a common feature representation to build the classifier ensemble
for questions having no labelled data. The proposed technique handsomely beats
all winning supervised entries on the SCIENTSBANK dataset from the Student
Response Analysis task of SemEval 2013. Additionally, we demonstrate
generalizability and benefits of the proposed technique through evaluation on
multiple ASAG datasets from different subject topics and standards.
| no_new_dataset | 0.947672 |
1609.08039 | Evgeny Burnaev | Evgeny Burnaev and Dmitry Smolyakov | One-Class SVM with Privileged Information and its Application to Malware
Detection | 8 pages, 5 figures, 3 tables | null | null | null | stat.ML cs.CR stat.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A number of important applied problems in engineering, finance and medicine
can be formulated as a problem of anomaly detection. A classical approach to
the problem is to describe a normal state using a one-class support vector
machine. Then to detect anomalies we quantify a distance from a new observation
to the constructed description of the normal class. In this paper we present a
new approach to the one-class classification. We formulate a new problem
statement and a corresponding algorithm that allow taking into account a
privileged information during the training phase. We evaluate performance of
the proposed approach using a synthetic dataset, as well as the publicly
available Microsoft Malware Classification Challenge dataset.
| [
{
"version": "v1",
"created": "Mon, 26 Sep 2016 16:01:02 GMT"
},
{
"version": "v2",
"created": "Sun, 20 Nov 2016 16:31:46 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Burnaev",
"Evgeny",
""
],
[
"Smolyakov",
"Dmitry",
""
]
] | TITLE: One-Class SVM with Privileged Information and its Application to Malware
Detection
ABSTRACT: A number of important applied problems in engineering, finance and medicine
can be formulated as a problem of anomaly detection. A classical approach to
the problem is to describe a normal state using a one-class support vector
machine. Then to detect anomalies we quantify a distance from a new observation
to the constructed description of the normal class. In this paper we present a
new approach to the one-class classification. We formulate a new problem
statement and a corresponding algorithm that allow taking into account a
privileged information during the training phase. We evaluate performance of
the proposed approach using a synthetic dataset, as well as the publicly
available Microsoft Malware Classification Challenge dataset.
| no_new_dataset | 0.944022 |
1611.05546 | Damien Teney | Damien Teney, Anton van den Hengel | Zero-Shot Visual Question Answering | null | null | null | null | cs.CV cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Part of the appeal of Visual Question Answering (VQA) is its promise to
answer new questions about previously unseen images. Most current methods
demand training questions that illustrate every possible concept, and will
therefore never achieve this capability, since the volume of required training
data would be prohibitive. Answering general questions about images requires
methods capable of Zero-Shot VQA, that is, methods able to answer questions
beyond the scope of the training questions. We propose a new evaluation
protocol for VQA methods which measures their ability to perform Zero-Shot VQA,
and in doing so highlights significant practical deficiencies of current
approaches, some of which are masked by the biases in current datasets. We
propose and evaluate several strategies for achieving Zero-Shot VQA, including
methods based on pretrained word embeddings, object classifiers with semantic
embeddings, and test-time retrieval of example images. Our extensive
experiments are intended to serve as baselines for Zero-Shot VQA, and they also
achieve state-of-the-art performance in the standard VQA evaluation setting.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 03:21:00 GMT"
},
{
"version": "v2",
"created": "Sun, 20 Nov 2016 21:51:24 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Teney",
"Damien",
""
],
[
"Hengel",
"Anton van den",
""
]
] | TITLE: Zero-Shot Visual Question Answering
ABSTRACT: Part of the appeal of Visual Question Answering (VQA) is its promise to
answer new questions about previously unseen images. Most current methods
demand training questions that illustrate every possible concept, and will
therefore never achieve this capability, since the volume of required training
data would be prohibitive. Answering general questions about images requires
methods capable of Zero-Shot VQA, that is, methods able to answer questions
beyond the scope of the training questions. We propose a new evaluation
protocol for VQA methods which measures their ability to perform Zero-Shot VQA,
and in doing so highlights significant practical deficiencies of current
approaches, some of which are masked by the biases in current datasets. We
propose and evaluate several strategies for achieving Zero-Shot VQA, including
methods based on pretrained word embeddings, object classifiers with semantic
embeddings, and test-time retrieval of example images. Our extensive
experiments are intended to serve as baselines for Zero-Shot VQA, and they also
achieve state-of-the-art performance in the standard VQA evaluation setting.
| no_new_dataset | 0.941169 |
1611.06362 | Hong Liu | Hong Liu, Rongrong Ji, Yongjian Wu, Feiyue Huang | Ordinal Constrained Binary Code Learning for Nearest Neighbor Search | Accepted to AAAI 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent years have witnessed extensive attention in binary code learning,
a.k.a. hashing, for nearest neighbor search problems. It has been seen that
high-dimensional data points can be quantized into binary codes to give an
efficient similarity approximation via Hamming distance. Among existing
schemes, ranking-based hashing is recent promising that targets at preserving
ordinal relations of ranking in the Hamming space to minimize retrieval loss.
However, the size of the ranking tuples, which shows the ordinal relations, is
quadratic or cubic to the size of training samples. By given a large-scale
training data set, it is very expensive to embed such ranking tuples in binary
code learning. Besides, it remains a dificulty to build ranking tuples
efficiently for most ranking-preserving hashing, which are deployed over an
ordinal graph-based setting. To handle these problems, we propose a novel
ranking-preserving hashing method, dubbed Ordinal Constraint Hashing (OCH),
which efficiently learns the optimal hashing functions with a graph-based
approximation to embed the ordinal relations. The core idea is to reduce the
size of ordinal graph with ordinal constraint projection, which preserves the
ordinal relations through a small data set (such as clusters or random
samples). In particular, to learn such hash functions effectively, we further
relax the discrete constraints and design a specific stochastic gradient decent
algorithm for optimization. Experimental results on three large-scale visual
search benchmark datasets, i.e. LabelMe, Tiny100K and GIST1M, show that the
proposed OCH method can achieve superior performance over the state-of-the-arts
approaches.
| [
{
"version": "v1",
"created": "Sat, 19 Nov 2016 13:24:10 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Liu",
"Hong",
""
],
[
"Ji",
"Rongrong",
""
],
[
"Wu",
"Yongjian",
""
],
[
"Huang",
"Feiyue",
""
]
] | TITLE: Ordinal Constrained Binary Code Learning for Nearest Neighbor Search
ABSTRACT: Recent years have witnessed extensive attention in binary code learning,
a.k.a. hashing, for nearest neighbor search problems. It has been seen that
high-dimensional data points can be quantized into binary codes to give an
efficient similarity approximation via Hamming distance. Among existing
schemes, ranking-based hashing is recent promising that targets at preserving
ordinal relations of ranking in the Hamming space to minimize retrieval loss.
However, the size of the ranking tuples, which shows the ordinal relations, is
quadratic or cubic to the size of training samples. By given a large-scale
training data set, it is very expensive to embed such ranking tuples in binary
code learning. Besides, it remains a dificulty to build ranking tuples
efficiently for most ranking-preserving hashing, which are deployed over an
ordinal graph-based setting. To handle these problems, we propose a novel
ranking-preserving hashing method, dubbed Ordinal Constraint Hashing (OCH),
which efficiently learns the optimal hashing functions with a graph-based
approximation to embed the ordinal relations. The core idea is to reduce the
size of ordinal graph with ordinal constraint projection, which preserves the
ordinal relations through a small data set (such as clusters or random
samples). In particular, to learn such hash functions effectively, we further
relax the discrete constraints and design a specific stochastic gradient decent
algorithm for optimization. Experimental results on three large-scale visual
search benchmark datasets, i.e. LabelMe, Tiny100K and GIST1M, show that the
proposed OCH method can achieve superior performance over the state-of-the-arts
approaches.
| no_new_dataset | 0.950915 |
1611.06495 | Jiawei Zhang | Jiawei Zhang, Jinshan Pan, Wei-Sheng Lai, Rynson Lau, Ming-Hsuan Yang | Learning Fully Convolutional Networks for Iterative Non-blind
Deconvolution | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose a fully convolutional networks for iterative
non-blind deconvolution We decompose the non-blind deconvolution problem into
image denoising and image deconvolution. We train a FCNN to remove noises in
the gradient domain and use the learned gradients to guide the image
deconvolution step. In contrast to the existing deep neural network based
methods, we iteratively deconvolve the blurred images in a multi-stage
framework. The proposed method is able to learn an adaptive image prior, which
keeps both local (details) and global (structures) information. Both
quantitative and qualitative evaluations on benchmark datasets demonstrate that
the proposed method performs favorably against state-of-the-art algorithms in
terms of quality and speed.
| [
{
"version": "v1",
"created": "Sun, 20 Nov 2016 10:25:06 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Zhang",
"Jiawei",
""
],
[
"Pan",
"Jinshan",
""
],
[
"Lai",
"Wei-Sheng",
""
],
[
"Lau",
"Rynson",
""
],
[
"Yang",
"Ming-Hsuan",
""
]
] | TITLE: Learning Fully Convolutional Networks for Iterative Non-blind
Deconvolution
ABSTRACT: In this paper, we propose a fully convolutional networks for iterative
non-blind deconvolution We decompose the non-blind deconvolution problem into
image denoising and image deconvolution. We train a FCNN to remove noises in
the gradient domain and use the learned gradients to guide the image
deconvolution step. In contrast to the existing deep neural network based
methods, we iteratively deconvolve the blurred images in a multi-stage
framework. The proposed method is able to learn an adaptive image prior, which
keeps both local (details) and global (structures) information. Both
quantitative and qualitative evaluations on benchmark datasets demonstrate that
the proposed method performs favorably against state-of-the-art algorithms in
terms of quality and speed.
| no_new_dataset | 0.947381 |
1611.06539 | Sebastian Vogel | Sebastian Vogel, Christoph Schorn, Andre Guntoro, Gerd Ascheid | Efficient Stochastic Inference of Bitwise Deep Neural Networks | 6 pages, 3 figures, Workshop on Efficient Methods for Deep Neural
Networks at Neural Information Processing Systems Conference 2016, NIPS 2016,
EMDNN 2016 | null | null | null | cs.NE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently published methods enable training of bitwise neural networks which
allow reduced representation of down to a single bit per weight. We present a
method that exploits ensemble decisions based on multiple stochastically
sampled network models to increase performance figures of bitwise neural
networks in terms of classification accuracy at inference. Our experiments with
the CIFAR-10 and GTSRB datasets show that the performance of such network
ensembles surpasses the performance of the high-precision base model. With this
technique we achieve 5.81% best classification error on CIFAR-10 test set using
bitwise networks. Concerning inference on embedded systems we evaluate these
bitwise networks using a hardware efficient stochastic rounding procedure. Our
work contributes to efficient embedded bitwise neural networks.
| [
{
"version": "v1",
"created": "Sun, 20 Nov 2016 16:05:07 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Vogel",
"Sebastian",
""
],
[
"Schorn",
"Christoph",
""
],
[
"Guntoro",
"Andre",
""
],
[
"Ascheid",
"Gerd",
""
]
] | TITLE: Efficient Stochastic Inference of Bitwise Deep Neural Networks
ABSTRACT: Recently published methods enable training of bitwise neural networks which
allow reduced representation of down to a single bit per weight. We present a
method that exploits ensemble decisions based on multiple stochastically
sampled network models to increase performance figures of bitwise neural
networks in terms of classification accuracy at inference. Our experiments with
the CIFAR-10 and GTSRB datasets show that the performance of such network
ensembles surpasses the performance of the high-precision base model. With this
technique we achieve 5.81% best classification error on CIFAR-10 test set using
bitwise networks. Concerning inference on embedded systems we evaluate these
bitwise networks using a hardware efficient stochastic rounding procedure. Our
work contributes to efficient embedded bitwise neural networks.
| no_new_dataset | 0.949763 |
1611.06625 | Geli Fei | Huayi Li, Geli Fei, Shuai Wang, Bing Liu, Weixiang Shao, Arjun
Mukherjee, Jidong Shao | Modeling Review Spam Using Temporal Patterns and Co-bursting Behaviors | null | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Online reviews play a crucial role in helping consumers evaluate and compare
products and services. However, review hosting sites are often targeted by
opinion spamming. In recent years, many such sites have put a great deal of
effort in building effective review filtering systems to detect fake reviews
and to block malicious accounts. Thus, fraudsters or spammers now turn to
compromise, purchase or even raise reputable accounts to write fake reviews.
Based on the analysis of a real-life dataset from a review hosting site
(dianping.com), we discovered that reviewers' posting rates are bimodal and the
transitions between different states can be utilized to differentiate spammers
from genuine reviewers. Inspired by these findings, we propose a two-mode
Labeled Hidden Markov Model to detect spammers. Experimental results show that
our model significantly outperforms supervised learning using linguistic and
behavioral features in identifying spammers. Furthermore, we found that when a
product has a burst of reviews, many spammers are likely to be actively
involved in writing reviews to the product as well as to many other products.
We then propose a novel co-bursting network for detecting spammer groups. The
co-bursting network enables us to produce more accurate spammer groups than the
current state-of-the-art reviewer-product (co-reviewing) network.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 01:13:42 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Li",
"Huayi",
""
],
[
"Fei",
"Geli",
""
],
[
"Wang",
"Shuai",
""
],
[
"Liu",
"Bing",
""
],
[
"Shao",
"Weixiang",
""
],
[
"Mukherjee",
"Arjun",
""
],
[
"Shao",
"Jidong",
""
]
] | TITLE: Modeling Review Spam Using Temporal Patterns and Co-bursting Behaviors
ABSTRACT: Online reviews play a crucial role in helping consumers evaluate and compare
products and services. However, review hosting sites are often targeted by
opinion spamming. In recent years, many such sites have put a great deal of
effort in building effective review filtering systems to detect fake reviews
and to block malicious accounts. Thus, fraudsters or spammers now turn to
compromise, purchase or even raise reputable accounts to write fake reviews.
Based on the analysis of a real-life dataset from a review hosting site
(dianping.com), we discovered that reviewers' posting rates are bimodal and the
transitions between different states can be utilized to differentiate spammers
from genuine reviewers. Inspired by these findings, we propose a two-mode
Labeled Hidden Markov Model to detect spammers. Experimental results show that
our model significantly outperforms supervised learning using linguistic and
behavioral features in identifying spammers. Furthermore, we found that when a
product has a burst of reviews, many spammers are likely to be actively
involved in writing reviews to the product as well as to many other products.
We then propose a novel co-bursting network for detecting spammer groups. The
co-bursting network enables us to produce more accurate spammer groups than the
current state-of-the-art reviewer-product (co-reviewing) network.
| no_new_dataset | 0.946399 |
1611.06638 | Qiang Qiu | Jose Lezama, Qiang Qiu, Guillermo Sapiro | Not Afraid of the Dark: NIR-VIS Face Recognition via Cross-spectral
Hallucination and Low-rank Embedding | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Surveillance cameras today often capture NIR (near infrared) images in
low-light environments. However, most face datasets accessible for training and
verification are only collected in the VIS (visible light) spectrum. It remains
a challenging problem to match NIR to VIS face images due to the different
light spectrum. Recently, breakthroughs have been made for VIS face recognition
by applying deep learning on a huge amount of labeled VIS face samples. The
same deep learning approach cannot be simply applied to NIR face recognition
for two main reasons: First, much limited NIR face images are available for
training compared to the VIS spectrum. Second, face galleries to be matched are
mostly available only in the VIS spectrum. In this paper, we propose an
approach to extend the deep learning breakthrough for VIS face recognition to
the NIR spectrum, without retraining the underlying deep models that see only
VIS faces. Our approach consists of two core components, cross-spectral
hallucination and low-rank embedding, to optimize respectively input and output
of a VIS deep model for cross-spectral face recognition. Cross-spectral
hallucination produces VIS faces from NIR images through a deep learning
approach. Low-rank embedding restores a low-rank structure for faces deep
features across both NIR and VIS spectrum. We observe that it is often equally
effective to perform hallucination to input NIR images or low-rank embedding to
output deep features for a VIS deep model for cross-spectral recognition. When
hallucination and low-rank embedding are deployed together, we observe
significant further improvement; we obtain state-of-the-art accuracy on the
CASIA NIR-VIS v2.0 benchmark, without the need at all to re-train the
recognition system.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 03:22:23 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Lezama",
"Jose",
""
],
[
"Qiu",
"Qiang",
""
],
[
"Sapiro",
"Guillermo",
""
]
] | TITLE: Not Afraid of the Dark: NIR-VIS Face Recognition via Cross-spectral
Hallucination and Low-rank Embedding
ABSTRACT: Surveillance cameras today often capture NIR (near infrared) images in
low-light environments. However, most face datasets accessible for training and
verification are only collected in the VIS (visible light) spectrum. It remains
a challenging problem to match NIR to VIS face images due to the different
light spectrum. Recently, breakthroughs have been made for VIS face recognition
by applying deep learning on a huge amount of labeled VIS face samples. The
same deep learning approach cannot be simply applied to NIR face recognition
for two main reasons: First, much limited NIR face images are available for
training compared to the VIS spectrum. Second, face galleries to be matched are
mostly available only in the VIS spectrum. In this paper, we propose an
approach to extend the deep learning breakthrough for VIS face recognition to
the NIR spectrum, without retraining the underlying deep models that see only
VIS faces. Our approach consists of two core components, cross-spectral
hallucination and low-rank embedding, to optimize respectively input and output
of a VIS deep model for cross-spectral face recognition. Cross-spectral
hallucination produces VIS faces from NIR images through a deep learning
approach. Low-rank embedding restores a low-rank structure for faces deep
features across both NIR and VIS spectrum. We observe that it is often equally
effective to perform hallucination to input NIR images or low-rank embedding to
output deep features for a VIS deep model for cross-spectral recognition. When
hallucination and low-rank embedding are deployed together, we observe
significant further improvement; we obtain state-of-the-art accuracy on the
CASIA NIR-VIS v2.0 benchmark, without the need at all to re-train the
recognition system.
| no_new_dataset | 0.953319 |
1611.06671 | Mark Magumba | Mark Abraham Magumba, Peter Nabende | Ontology Driven Disease Incidence Detection on Twitter | 19 pages, 7 figures, 1 table | null | null | null | cs.CL cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we address the issue of generic automated disease incidence
monitoring on twitter. We employ an ontology of disease related concepts and
use it to obtain a conceptual representation of tweets. Unlike previous key
word based systems and topic modeling approaches, our ontological approach
allows us to apply more stringent criteria for determining which messages are
relevant such as spatial and temporal characteristics whilst giving a stronger
guarantee that the resulting models will perform well on new data that may be
lexically divergent. We achieve this by training learners on concepts rather
than individual words. For training we use a dataset containing mentions of
influenza and Listeria and use the learned models to classify datasets
containing mentions of an arbitrary selection of other diseases. We show that
our ontological approach achieves good performance on this task using a variety
of Natural Language Processing Techniques. We also show that word vectors can
be learned directly from our concepts to achieve even better results.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 07:32:56 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Magumba",
"Mark Abraham",
""
],
[
"Nabende",
"Peter",
""
]
] | TITLE: Ontology Driven Disease Incidence Detection on Twitter
ABSTRACT: In this work we address the issue of generic automated disease incidence
monitoring on twitter. We employ an ontology of disease related concepts and
use it to obtain a conceptual representation of tweets. Unlike previous key
word based systems and topic modeling approaches, our ontological approach
allows us to apply more stringent criteria for determining which messages are
relevant such as spatial and temporal characteristics whilst giving a stronger
guarantee that the resulting models will perform well on new data that may be
lexically divergent. We achieve this by training learners on concepts rather
than individual words. For training we use a dataset containing mentions of
influenza and Listeria and use the learned models to classify datasets
containing mentions of an arbitrary selection of other diseases. We show that
our ontological approach achieves good performance on this task using a variety
of Natural Language Processing Techniques. We also show that word vectors can
be learned directly from our concepts to achieve even better results.
| no_new_dataset | 0.940079 |
1611.06678 | Ali Diba | Ali Diba, Vivek Sharma, Luc Van Gool | Deep Temporal Linear Encoding Networks | Ali Diba and Vivek Sharma contributed equally to this work and listed
in alphabetical order | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The CNN-encoding of features from entire videos for the representation of
human actions has rarely been addressed. Instead, CNN work has focused on
approaches to fuse spatial and temporal networks, but these were typically
limited to processing shorter sequences. We present a new video representation,
called temporal linear encoding (TLE) and embedded inside of CNNs as a new
layer, which captures the appearance and motion throughout entire videos. It
encodes this aggregated information into a robust video feature representation,
via end-to-end learning. Advantages of TLEs are: (a) they encode the entire
video into a compact feature representation, learning the semantics and a
discriminative feature space; (b) they are applicable to all kinds of networks
like 2D and 3D CNNs for video classification; and (c) they model feature
interactions in a more expressive way and without loss of information. We
conduct experiments on two challenging human action datasets: HMDB51 and
UCF101. The experiments show that TLE outperforms current state-of-the-art
methods on both datasets.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 08:27:31 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Diba",
"Ali",
""
],
[
"Sharma",
"Vivek",
""
],
[
"Van Gool",
"Luc",
""
]
] | TITLE: Deep Temporal Linear Encoding Networks
ABSTRACT: The CNN-encoding of features from entire videos for the representation of
human actions has rarely been addressed. Instead, CNN work has focused on
approaches to fuse spatial and temporal networks, but these were typically
limited to processing shorter sequences. We present a new video representation,
called temporal linear encoding (TLE) and embedded inside of CNNs as a new
layer, which captures the appearance and motion throughout entire videos. It
encodes this aggregated information into a robust video feature representation,
via end-to-end learning. Advantages of TLEs are: (a) they encode the entire
video into a compact feature representation, learning the semantics and a
discriminative feature space; (b) they are applicable to all kinds of networks
like 2D and 3D CNNs for video classification; and (c) they model feature
interactions in a more expressive way and without loss of information. We
conduct experiments on two challenging human action datasets: HMDB51 and
UCF101. The experiments show that TLE outperforms current state-of-the-art
methods on both datasets.
| no_new_dataset | 0.948489 |
1611.06683 | HImanshu Aggarwal | Himanshu Aggarwal, Dinesh K. Vishwakarma | Covariate conscious approach for Gait recognition based upon Zernike
moment invariants | 11 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Gait recognition i.e. identification of an individual from his/her walking
pattern is an emerging field. While existing gait recognition techniques
perform satisfactorily in normal walking conditions, there performance tend to
suffer drastically with variations in clothing and carrying conditions. In this
work, we propose a novel covariate cognizant framework to deal with the
presence of such covariates. We describe gait motion by forming a single 2D
spatio-temporal template from video sequence, called Average Energy Silhouette
image (AESI). Zernike moment invariants (ZMIs) are then computed to screen the
parts of AESI infected with covariates. Following this, features are extracted
from Spatial Distribution of Oriented Gradients (SDOGs) and novel Mean of
Directional Pixels (MDPs) methods. The obtained features are fused together to
form the final well-endowed feature set. Experimental evaluation of the
proposed framework on three publicly available datasets i.e. CASIA dataset B,
OU-ISIR Treadmill dataset B and USF Human-ID challenge dataset with recently
published gait recognition approaches, prove its superior performance.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 08:48:47 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Aggarwal",
"Himanshu",
""
],
[
"Vishwakarma",
"Dinesh K.",
""
]
] | TITLE: Covariate conscious approach for Gait recognition based upon Zernike
moment invariants
ABSTRACT: Gait recognition i.e. identification of an individual from his/her walking
pattern is an emerging field. While existing gait recognition techniques
perform satisfactorily in normal walking conditions, there performance tend to
suffer drastically with variations in clothing and carrying conditions. In this
work, we propose a novel covariate cognizant framework to deal with the
presence of such covariates. We describe gait motion by forming a single 2D
spatio-temporal template from video sequence, called Average Energy Silhouette
image (AESI). Zernike moment invariants (ZMIs) are then computed to screen the
parts of AESI infected with covariates. Following this, features are extracted
from Spatial Distribution of Oriented Gradients (SDOGs) and novel Mean of
Directional Pixels (MDPs) methods. The obtained features are fused together to
form the final well-endowed feature set. Experimental evaluation of the
proposed framework on three publicly available datasets i.e. CASIA dataset B,
OU-ISIR Treadmill dataset B and USF Human-ID challenge dataset with recently
published gait recognition approaches, prove its superior performance.
| no_new_dataset | 0.935993 |
1611.06748 | Di Kang | Di Kang, Debarun Dhar, Antoni B. Chan | Crowd Counting by Adapting Convolutional Neural Networks with Side
Information | 8 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Computer vision tasks often have side information available that is helpful
to solve the task. For example, for crowd counting, the camera perspective
(e.g., camera angle and height) gives a clue about the appearance and scale of
people in the scene. While side information has been shown to be useful for
counting systems using traditional hand-crafted features, it has not been fully
utilized in counting systems based on deep learning. In order to incorporate
the available side information, we propose an adaptive convolutional neural
network (ACNN), where the convolutional filter weights adapt to the current
scene context via the side information. In particular, we model the filter
weights as a low-dimensional manifold, parametrized by the side information,
within the high-dimensional space of filter weights. With the help of side
information and adaptive weights, the ACNN can disentangle the variations
related to the side information, and extract discriminative features related to
the current context. Since existing crowd counting datasets do not contain
ground-truth side information, we collect a new dataset with the ground-truth
camera angle and height as the side information. On experiments in crowd
counting, the ACNN improves counting accuracy compared to a plain CNN with a
similar number of parameters. We also apply ACNN to image deconvolution to show
its potential effectiveness on other computer vision applications.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 12:09:06 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Kang",
"Di",
""
],
[
"Dhar",
"Debarun",
""
],
[
"Chan",
"Antoni B.",
""
]
] | TITLE: Crowd Counting by Adapting Convolutional Neural Networks with Side
Information
ABSTRACT: Computer vision tasks often have side information available that is helpful
to solve the task. For example, for crowd counting, the camera perspective
(e.g., camera angle and height) gives a clue about the appearance and scale of
people in the scene. While side information has been shown to be useful for
counting systems using traditional hand-crafted features, it has not been fully
utilized in counting systems based on deep learning. In order to incorporate
the available side information, we propose an adaptive convolutional neural
network (ACNN), where the convolutional filter weights adapt to the current
scene context via the side information. In particular, we model the filter
weights as a low-dimensional manifold, parametrized by the side information,
within the high-dimensional space of filter weights. With the help of side
information and adaptive weights, the ACNN can disentangle the variations
related to the side information, and extract discriminative features related to
the current context. Since existing crowd counting datasets do not contain
ground-truth side information, we collect a new dataset with the ground-truth
camera angle and height as the side information. On experiments in crowd
counting, the ACNN improves counting accuracy compared to a plain CNN with a
similar number of parameters. We also apply ACNN to image deconvolution to show
its potential effectiveness on other computer vision applications.
| new_dataset | 0.952175 |
1611.06969 | Raphael Felipe Prates | Raphael Prates and William Robson Schwartz | Kernel Cross-View Collaborative Representation based Classification for
Person Re-Identification | Paper submitted to CVPR 2017 conference | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Person re-identification aims at the maintenance of a global identity as a
person moves among non-overlapping surveillance cameras. It is a hard task due
to different illumination conditions, viewpoints and the small number of
annotated individuals from each pair of cameras (small-sample-size problem).
Collaborative Representation based Classification (CRC) has been employed
successfully to address the small-sample-size problem in computer vision.
However, the original CRC formulation is not well-suited for person
re-identification since it does not consider that probe and gallery samples are
from different cameras. Furthermore, it is a linear model, while appearance
changes caused by different camera conditions indicate a strong nonlinear
transition between cameras. To overcome such limitations, we propose the Kernel
Cross-View Collaborative Representation based Classification (Kernel X-CRC)
that represents probe and gallery images by balancing representativeness and
similarity nonlinearly. It assumes that a probe and its corresponding gallery
image are represented with similar coding vectors using individuals from the
training set. Experimental results demonstrate that our assumption is true when
using a high-dimensional feature vector and becomes more compelling when
dealing with a low-dimensional and discriminative representation computed using
a common subspace learning method. We achieve state-of-the-art for rank-1
matching rates in two person re-identification datasets (PRID450S and GRID) and
the second best results on VIPeR and CUHK01 datasets.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 19:44:50 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Prates",
"Raphael",
""
],
[
"Schwartz",
"William Robson",
""
]
] | TITLE: Kernel Cross-View Collaborative Representation based Classification for
Person Re-Identification
ABSTRACT: Person re-identification aims at the maintenance of a global identity as a
person moves among non-overlapping surveillance cameras. It is a hard task due
to different illumination conditions, viewpoints and the small number of
annotated individuals from each pair of cameras (small-sample-size problem).
Collaborative Representation based Classification (CRC) has been employed
successfully to address the small-sample-size problem in computer vision.
However, the original CRC formulation is not well-suited for person
re-identification since it does not consider that probe and gallery samples are
from different cameras. Furthermore, it is a linear model, while appearance
changes caused by different camera conditions indicate a strong nonlinear
transition between cameras. To overcome such limitations, we propose the Kernel
Cross-View Collaborative Representation based Classification (Kernel X-CRC)
that represents probe and gallery images by balancing representativeness and
similarity nonlinearly. It assumes that a probe and its corresponding gallery
image are represented with similar coding vectors using individuals from the
training set. Experimental results demonstrate that our assumption is true when
using a high-dimensional feature vector and becomes more compelling when
dealing with a low-dimensional and discriminative representation computed using
a common subspace learning method. We achieve state-of-the-art for rank-1
matching rates in two person re-identification datasets (PRID450S and GRID) and
the second best results on VIPeR and CUHK01 datasets.
| no_new_dataset | 0.953057 |
1611.06973 | Seymour Knowles-Barley | Seymour Knowles-Barley, Verena Kaynig, Thouis Ray Jones, Alyssa
Wilson, Joshua Morgan, Dongil Lee, Daniel Berger, Narayanan Kasthuri, Jeff W.
Lichtman, Hanspeter Pfister | RhoanaNet Pipeline: Dense Automatic Neural Annotation | 13 pages, 4 figures | null | null | null | q-bio.NC cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reconstructing a synaptic wiring diagram, or connectome, from electron
microscopy (EM) images of brain tissue currently requires many hours of manual
annotation or proofreading (Kasthuri and Lichtman, 2010; Lichtman and Sanes,
2008; Seung, 2009). The desire to reconstruct ever larger and more complex
networks has pushed the collection of ever larger EM datasets. A cubic
millimeter of raw imaging data would take up 1 PB of storage and present an
annotation project that would be impractical without relying heavily on
automatic segmentation methods. The RhoanaNet image processing pipeline was
developed to automatically segment large volumes of EM data and ease the burden
of manual proofreading and annotation. Based on (Kaynig et al., 2015), we
updated every stage of the software pipeline to provide better throughput
performance and higher quality segmentation results. We used state of the art
deep learning techniques to generate improved membrane probability maps, and
Gala (Nunez-Iglesias et al., 2014) was used to agglomerate 2D segments into 3D
objects.
We applied the RhoanaNet pipeline to four densely annotated EM datasets, two
from mouse cortex, one from cerebellum and one from mouse lateral geniculate
nucleus (LGN). All training and test data is made available for benchmark
comparisons. The best segmentation results obtained gave
$V^\text{Info}_\text{F-score}$ scores of 0.9054 and 09182 for the cortex
datasets, 0.9438 for LGN, and 0.9150 for Cerebellum.
The RhoanaNet pipeline is open source software. All source code, training
data, test data, and annotations for all four benchmark datasets are available
at www.rhoana.org.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 19:48:29 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Knowles-Barley",
"Seymour",
""
],
[
"Kaynig",
"Verena",
""
],
[
"Jones",
"Thouis Ray",
""
],
[
"Wilson",
"Alyssa",
""
],
[
"Morgan",
"Joshua",
""
],
[
"Lee",
"Dongil",
""
],
[
"Berger",
"Daniel",
""
],
[
"Kasthuri",
"Narayanan",
""
],
[
"Lichtman",
"Jeff W.",
""
],
[
"Pfister",
"Hanspeter",
""
]
] | TITLE: RhoanaNet Pipeline: Dense Automatic Neural Annotation
ABSTRACT: Reconstructing a synaptic wiring diagram, or connectome, from electron
microscopy (EM) images of brain tissue currently requires many hours of manual
annotation or proofreading (Kasthuri and Lichtman, 2010; Lichtman and Sanes,
2008; Seung, 2009). The desire to reconstruct ever larger and more complex
networks has pushed the collection of ever larger EM datasets. A cubic
millimeter of raw imaging data would take up 1 PB of storage and present an
annotation project that would be impractical without relying heavily on
automatic segmentation methods. The RhoanaNet image processing pipeline was
developed to automatically segment large volumes of EM data and ease the burden
of manual proofreading and annotation. Based on (Kaynig et al., 2015), we
updated every stage of the software pipeline to provide better throughput
performance and higher quality segmentation results. We used state of the art
deep learning techniques to generate improved membrane probability maps, and
Gala (Nunez-Iglesias et al., 2014) was used to agglomerate 2D segments into 3D
objects.
We applied the RhoanaNet pipeline to four densely annotated EM datasets, two
from mouse cortex, one from cerebellum and one from mouse lateral geniculate
nucleus (LGN). All training and test data is made available for benchmark
comparisons. The best segmentation results obtained gave
$V^\text{Info}_\text{F-score}$ scores of 0.9054 and 09182 for the cortex
datasets, 0.9438 for LGN, and 0.9150 for Cerebellum.
The RhoanaNet pipeline is open source software. All source code, training
data, test data, and annotations for all four benchmark datasets are available
at www.rhoana.org.
| no_new_dataset | 0.951549 |
1611.06997 | Hongyuan Mei | Hongyuan Mei and Mohit Bansal and Matthew R. Walter | Coherent Dialogue with Attention-based Language Models | To appear at AAAI 2017 | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We model coherent conversation continuation via RNN-based dialogue models
equipped with a dynamic attention mechanism. Our attention-RNN language model
dynamically increases the scope of attention on the history as the conversation
continues, as opposed to standard attention (or alignment) models with a fixed
input scope in a sequence-to-sequence model. This allows each generated word to
be associated with the most relevant words in its corresponding conversation
history. We evaluate the model on two popular dialogue datasets, the
open-domain MovieTriples dataset and the closed-domain Ubuntu Troubleshoot
dataset, and achieve significant improvements over the state-of-the-art and
baselines on several metrics, including complementary diversity-based metrics,
human evaluation, and qualitative visualizations. We also show that a vanilla
RNN with dynamic attention outperforms more complex memory models (e.g., LSTM
and GRU) by allowing for flexible, long-distance memory. We promote further
coherence via topic modeling-based reranking.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 20:25:19 GMT"
}
] | 2016-11-22T00:00:00 | [
[
"Mei",
"Hongyuan",
""
],
[
"Bansal",
"Mohit",
""
],
[
"Walter",
"Matthew R.",
""
]
] | TITLE: Coherent Dialogue with Attention-based Language Models
ABSTRACT: We model coherent conversation continuation via RNN-based dialogue models
equipped with a dynamic attention mechanism. Our attention-RNN language model
dynamically increases the scope of attention on the history as the conversation
continues, as opposed to standard attention (or alignment) models with a fixed
input scope in a sequence-to-sequence model. This allows each generated word to
be associated with the most relevant words in its corresponding conversation
history. We evaluate the model on two popular dialogue datasets, the
open-domain MovieTriples dataset and the closed-domain Ubuntu Troubleshoot
dataset, and achieve significant improvements over the state-of-the-art and
baselines on several metrics, including complementary diversity-based metrics,
human evaluation, and qualitative visualizations. We also show that a vanilla
RNN with dynamic attention outperforms more complex memory models (e.g., LSTM
and GRU) by allowing for flexible, long-distance memory. We promote further
coherence via topic modeling-based reranking.
| no_new_dataset | 0.950134 |
1506.06318 | Shang-Tse Chen | Shang-Tse Chen, Maria-Florina Balcan, Duen Horng Chau | Communication Efficient Distributed Agnostic Boosting | null | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the problem of learning from distributed data in the agnostic
setting, i.e., in the presence of arbitrary forms of noise. Our main
contribution is a general distributed boosting-based procedure for learning an
arbitrary concept space, that is simultaneously noise tolerant, communication
efficient, and computationally efficient. This improves significantly over
prior works that were either communication efficient only in noise-free
scenarios or computationally prohibitive. Empirical results on large synthetic
and real-world datasets demonstrate the effectiveness and scalability of the
proposed approach.
| [
{
"version": "v1",
"created": "Sun, 21 Jun 2015 04:35:42 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Nov 2016 16:55:03 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Chen",
"Shang-Tse",
""
],
[
"Balcan",
"Maria-Florina",
""
],
[
"Chau",
"Duen Horng",
""
]
] | TITLE: Communication Efficient Distributed Agnostic Boosting
ABSTRACT: We consider the problem of learning from distributed data in the agnostic
setting, i.e., in the presence of arbitrary forms of noise. Our main
contribution is a general distributed boosting-based procedure for learning an
arbitrary concept space, that is simultaneously noise tolerant, communication
efficient, and computationally efficient. This improves significantly over
prior works that were either communication efficient only in noise-free
scenarios or computationally prohibitive. Empirical results on large synthetic
and real-world datasets demonstrate the effectiveness and scalability of the
proposed approach.
| no_new_dataset | 0.950041 |
1510.08431 | Christopher Grant | C. Grant and B. R. Littlejohn | Opportunities With Decay-At-Rest Neutrinos From Decay-In-Flight Neutrino
Beams | 6 pages, 3 figures | null | null | null | hep-ex hep-ph nucl-ex physics.ins-det | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Neutrino beam facilities, like spallation neutron facilities, produce copious
quantities of neutrinos from the decay at rest of mesons and muons. The
viability of decay-in-flight neutrino beams as sites for decay-at-rest neutrino
studies has been investigated by calculating expected low-energy neutrino
fluxes from the existing Fermilab NuMI beam facility. Decay-at-rest neutrino
production in NuMI is found to be roughly equivalent per megawatt to that of
spallation facilities, and is concentrated in the facility's target hall and
beam stop regions. Interaction rates in 5 and 60 ton liquid argon detectors at
a variety of existing and hypothetical locations along the beamline are found
to be comparable to the largest existing decay-at-rest datasets for some
channels. The physics implications and experimental challenges of such a
measurement are discussed, along with prospects for measurements at targeted
facilities along a future Fermilab long-baseline neutrino beam.
| [
{
"version": "v1",
"created": "Wed, 28 Oct 2015 19:37:10 GMT"
},
{
"version": "v2",
"created": "Mon, 16 Nov 2015 20:13:59 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Nov 2016 22:23:06 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Grant",
"C.",
""
],
[
"Littlejohn",
"B. R.",
""
]
] | TITLE: Opportunities With Decay-At-Rest Neutrinos From Decay-In-Flight Neutrino
Beams
ABSTRACT: Neutrino beam facilities, like spallation neutron facilities, produce copious
quantities of neutrinos from the decay at rest of mesons and muons. The
viability of decay-in-flight neutrino beams as sites for decay-at-rest neutrino
studies has been investigated by calculating expected low-energy neutrino
fluxes from the existing Fermilab NuMI beam facility. Decay-at-rest neutrino
production in NuMI is found to be roughly equivalent per megawatt to that of
spallation facilities, and is concentrated in the facility's target hall and
beam stop regions. Interaction rates in 5 and 60 ton liquid argon detectors at
a variety of existing and hypothetical locations along the beamline are found
to be comparable to the largest existing decay-at-rest datasets for some
channels. The physics implications and experimental challenges of such a
measurement are discussed, along with prospects for measurements at targeted
facilities along a future Fermilab long-baseline neutrino beam.
| no_new_dataset | 0.939137 |
1511.03719 | Xiang Zhang | Xiang Zhang, Yann LeCun | Universum Prescription: Regularization using Unlabeled Data | 7 pages for article, 3 pages for supplemental material. To appear in
AAAI-17 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper shows that simply prescribing "none of the above" labels to
unlabeled data has a beneficial regularization effect to supervised learning.
We call it universum prescription by the fact that the prescribed labels cannot
be one of the supervised labels. In spite of its simplicity, universum
prescription obtained competitive results in training deep convolutional
networks for CIFAR-10, CIFAR-100, STL-10 and ImageNet datasets. A qualitative
justification of these approaches using Rademacher complexity is presented. The
effect of a regularization parameter -- probability of sampling from unlabeled
data -- is also studied empirically.
| [
{
"version": "v1",
"created": "Wed, 11 Nov 2015 22:46:46 GMT"
},
{
"version": "v2",
"created": "Wed, 18 Nov 2015 19:54:22 GMT"
},
{
"version": "v3",
"created": "Sun, 22 Nov 2015 22:12:09 GMT"
},
{
"version": "v4",
"created": "Thu, 21 Jan 2016 06:11:33 GMT"
},
{
"version": "v5",
"created": "Mon, 15 Feb 2016 18:52:30 GMT"
},
{
"version": "v6",
"created": "Mon, 25 Apr 2016 21:10:32 GMT"
},
{
"version": "v7",
"created": "Fri, 18 Nov 2016 01:15:30 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Zhang",
"Xiang",
""
],
[
"LeCun",
"Yann",
""
]
] | TITLE: Universum Prescription: Regularization using Unlabeled Data
ABSTRACT: This paper shows that simply prescribing "none of the above" labels to
unlabeled data has a beneficial regularization effect to supervised learning.
We call it universum prescription by the fact that the prescribed labels cannot
be one of the supervised labels. In spite of its simplicity, universum
prescription obtained competitive results in training deep convolutional
networks for CIFAR-10, CIFAR-100, STL-10 and ImageNet datasets. A qualitative
justification of these approaches using Rademacher complexity is presented. The
effect of a regularization parameter -- probability of sampling from unlabeled
data -- is also studied empirically.
| no_new_dataset | 0.949389 |
1604.08354 | Anne-Florence Bitbol | Anne-Florence Bitbol, Robert S. Dwyer, Lucy J. Colwell and Ned S.
Wingreen | Inferring interaction partners from protein sequences | 25 pages, 19 figures, published version | Proc. Natl. Acad. Sci. U.S.A., 113(43): 12180-12185 (2016) | 10.1073/pnas.1606762113 | null | physics.bio-ph q-bio.BM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Specific protein-protein interactions are crucial in the cell, both to ensure
the formation and stability of multi-protein complexes, and to enable signal
transduction in various pathways. Functional interactions between proteins
result in coevolution between the interaction partners, causing their sequences
to be correlated. Here we exploit these correlations to accurately identify
which proteins are specific interaction partners from sequence data alone. Our
general approach, which employs a pairwise maximum entropy model to infer
couplings between residues, has been successfully used to predict the
three-dimensional structures of proteins from sequences. Thus inspired, we
introduce an iterative algorithm to predict specific interaction partners from
two protein families whose members are known to interact. We first assess the
algorithm's performance on histidine kinases and response regulators from
bacterial two-component signaling systems. We obtain a striking 0.93 true
positive fraction on our complete dataset without any a priori knowledge of
interaction partners, and we uncover the origin of this success. We then apply
the algorithm to proteins from ATP-binding cassette (ABC) transporter
complexes, and obtain accurate predictions in these systems as well. Finally,
we present two metrics that accurately distinguish interacting protein families
from non-interacting ones, using only sequence data.
| [
{
"version": "v1",
"created": "Thu, 28 Apr 2016 09:23:57 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Nov 2016 21:29:46 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Bitbol",
"Anne-Florence",
""
],
[
"Dwyer",
"Robert S.",
""
],
[
"Colwell",
"Lucy J.",
""
],
[
"Wingreen",
"Ned S.",
""
]
] | TITLE: Inferring interaction partners from protein sequences
ABSTRACT: Specific protein-protein interactions are crucial in the cell, both to ensure
the formation and stability of multi-protein complexes, and to enable signal
transduction in various pathways. Functional interactions between proteins
result in coevolution between the interaction partners, causing their sequences
to be correlated. Here we exploit these correlations to accurately identify
which proteins are specific interaction partners from sequence data alone. Our
general approach, which employs a pairwise maximum entropy model to infer
couplings between residues, has been successfully used to predict the
three-dimensional structures of proteins from sequences. Thus inspired, we
introduce an iterative algorithm to predict specific interaction partners from
two protein families whose members are known to interact. We first assess the
algorithm's performance on histidine kinases and response regulators from
bacterial two-component signaling systems. We obtain a striking 0.93 true
positive fraction on our complete dataset without any a priori knowledge of
interaction partners, and we uncover the origin of this success. We then apply
the algorithm to proteins from ATP-binding cassette (ABC) transporter
complexes, and obtain accurate predictions in these systems as well. Finally,
we present two metrics that accurately distinguish interacting protein families
from non-interacting ones, using only sequence data.
| no_new_dataset | 0.944791 |
1611.04023 | Gerard Rinkus | Gerard J. Rinkus | Sparsey: Event Recognition via Deep Hierarchical Spare Distributed Codes | This is a manuscript form of a paper published in Frontiers in
Computational Neuroscience in 2014
(http://dx.doi.org/10.3389/fncom.2014.00160). 65 pages, 28 figures, 8 tables | Frontiers in Computational Neuroscience, Vol. 8, Article 160
(2014) | 10.3389/fncom.2014.00160 | null | q-bio.NC cs.CV cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Visual cortex's hierarchical, multi-level organization is captured in many
biologically inspired computational vision models, the general idea being that
progressively larger scale, more complex spatiotemporal features are
represented in progressively higher areas. However, most earlier models use
localist representations (codes) in each representational field, which we
equate with the cortical macrocolumn (mac), at each level. In localism, each
represented feature/event (item) is coded by a single unit. Our model, Sparsey,
is also hierarchical but crucially, uses sparse distributed coding (SDC) in
every mac in all levels. In SDC, each represented item is coded by a small
subset of the mac's units. SDCs of different items can overlap and the size of
overlap between items can represent their similarity. The difference between
localism and SDC is crucial because SDC allows the two essential operations of
associative memory, storing a new item and retrieving the best-matching stored
item, to be done in fixed time for the life of the model. Since the model's
core algorithm, which does both storage and retrieval (inference), makes a
single pass over all macs on each time step, the overall model's
storage/retrieval operation is also fixed-time, a criterion we consider
essential for scalability to huge datasets. A 2010 paper described a
nonhierarchical version of this model in the context of purely spatial pattern
processing. Here, we elaborate a fully hierarchical model (arbitrary numbers of
levels and macs per level), describing novel model principles like progressive
critical periods, dynamic modulation of principal cells' activation functions
based on a mac-level familiarity measure, representation of multiple
simultaneously active hypotheses, a novel method of time warp invariant
recognition, and we report results showing learning/recognition of
spatiotemporal patterns.
| [
{
"version": "v1",
"created": "Sat, 12 Nov 2016 17:35:23 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Rinkus",
"Gerard J.",
""
]
] | TITLE: Sparsey: Event Recognition via Deep Hierarchical Spare Distributed Codes
ABSTRACT: Visual cortex's hierarchical, multi-level organization is captured in many
biologically inspired computational vision models, the general idea being that
progressively larger scale, more complex spatiotemporal features are
represented in progressively higher areas. However, most earlier models use
localist representations (codes) in each representational field, which we
equate with the cortical macrocolumn (mac), at each level. In localism, each
represented feature/event (item) is coded by a single unit. Our model, Sparsey,
is also hierarchical but crucially, uses sparse distributed coding (SDC) in
every mac in all levels. In SDC, each represented item is coded by a small
subset of the mac's units. SDCs of different items can overlap and the size of
overlap between items can represent their similarity. The difference between
localism and SDC is crucial because SDC allows the two essential operations of
associative memory, storing a new item and retrieving the best-matching stored
item, to be done in fixed time for the life of the model. Since the model's
core algorithm, which does both storage and retrieval (inference), makes a
single pass over all macs on each time step, the overall model's
storage/retrieval operation is also fixed-time, a criterion we consider
essential for scalability to huge datasets. A 2010 paper described a
nonhierarchical version of this model in the context of purely spatial pattern
processing. Here, we elaborate a fully hierarchical model (arbitrary numbers of
levels and macs per level), describing novel model principles like progressive
critical periods, dynamic modulation of principal cells' activation functions
based on a mac-level familiarity measure, representation of multiple
simultaneously active hypotheses, a novel method of time warp invariant
recognition, and we report results showing learning/recognition of
spatiotemporal patterns.
| no_new_dataset | 0.953013 |
1611.05520 | Mohammad Sadegh Aliakbarian | Mohammad Sadegh Aliakbarian, Fatemehsadat Saleh, Basura Fernando,
Mathieu Salzmann, Lars Petersson, Lars Andersson | Deep Action- and Context-Aware Sequence Learning for Activity
Recognition and Anticipation | 10 pages, 4 figures, 7 tables | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Action recognition and anticipation are key to the success of many computer
vision applications. Existing methods can roughly be grouped into those that
extract global, context-aware representations of the entire image or sequence,
and those that aim at focusing on the regions where the action occurs. While
the former may suffer from the fact that context is not always reliable, the
latter completely ignore this source of information, which can nonetheless be
helpful in many situations. In this paper, we aim at making the best of both
worlds by developing an approach that leverages both context-aware and
action-aware features. At the core of our method lies a novel multi-stage
recurrent architecture that allows us to effectively combine these two sources
of information throughout a video. This architecture first exploits the global,
context-aware features, and merges the resulting representation with the
localized, action-aware ones. Our experiments on standard datasets evidence the
benefits of our approach over methods that use each information type
separately. We outperform the state-of-the-art methods that, as us, rely only
on RGB frames as input for both action recognition and anticipation.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 01:08:56 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Nov 2016 01:41:40 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Aliakbarian",
"Mohammad Sadegh",
""
],
[
"Saleh",
"Fatemehsadat",
""
],
[
"Fernando",
"Basura",
""
],
[
"Salzmann",
"Mathieu",
""
],
[
"Petersson",
"Lars",
""
],
[
"Andersson",
"Lars",
""
]
] | TITLE: Deep Action- and Context-Aware Sequence Learning for Activity
Recognition and Anticipation
ABSTRACT: Action recognition and anticipation are key to the success of many computer
vision applications. Existing methods can roughly be grouped into those that
extract global, context-aware representations of the entire image or sequence,
and those that aim at focusing on the regions where the action occurs. While
the former may suffer from the fact that context is not always reliable, the
latter completely ignore this source of information, which can nonetheless be
helpful in many situations. In this paper, we aim at making the best of both
worlds by developing an approach that leverages both context-aware and
action-aware features. At the core of our method lies a novel multi-stage
recurrent architecture that allows us to effectively combine these two sources
of information throughout a video. This architecture first exploits the global,
context-aware features, and merges the resulting representation with the
localized, action-aware ones. Our experiments on standard datasets evidence the
benefits of our approach over methods that use each information type
separately. We outperform the state-of-the-art methods that, as us, rely only
on RGB frames as input for both action recognition and anticipation.
| no_new_dataset | 0.943608 |
1611.05896 | Somak Aditya | Somak Aditya, Yezhou Yang, Chitta Baral, Yiannis Aloimonos | Answering Image Riddles using Vision and Reasoning through Probabilistic
Soft Logic | 14 pages, 10 figures | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we explore a genre of puzzles ("image riddles") which involves
a set of images and a question. Answering these puzzles require both
capabilities involving visual detection (including object, activity
recognition) and, knowledge-based or commonsense reasoning. We compile a
dataset of over 3k riddles where each riddle consists of 4 images and a
groundtruth answer. The annotations are validated using crowd-sourced
evaluation. We also define an automatic evaluation metric to track future
progress. Our task bears similarity with the commonly known IQ tasks such as
analogy solving, sequence filling that are often used to test intelligence.
We develop a Probabilistic Reasoning-based approach that utilizes
probabilistic commonsense knowledge to answer these riddles with a reasonable
accuracy. We demonstrate the results of our approach using both automatic and
human evaluations. Our approach achieves some promising results for these
riddles and provides a strong baseline for future attempts. We make the entire
dataset and related materials publicly available to the community in
ImageRiddle Website (http://bit.ly/22f9Ala).
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 21:10:33 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Aditya",
"Somak",
""
],
[
"Yang",
"Yezhou",
""
],
[
"Baral",
"Chitta",
""
],
[
"Aloimonos",
"Yiannis",
""
]
] | TITLE: Answering Image Riddles using Vision and Reasoning through Probabilistic
Soft Logic
ABSTRACT: In this work, we explore a genre of puzzles ("image riddles") which involves
a set of images and a question. Answering these puzzles require both
capabilities involving visual detection (including object, activity
recognition) and, knowledge-based or commonsense reasoning. We compile a
dataset of over 3k riddles where each riddle consists of 4 images and a
groundtruth answer. The annotations are validated using crowd-sourced
evaluation. We also define an automatic evaluation metric to track future
progress. Our task bears similarity with the commonly known IQ tasks such as
analogy solving, sequence filling that are often used to test intelligence.
We develop a Probabilistic Reasoning-based approach that utilizes
probabilistic commonsense knowledge to answer these riddles with a reasonable
accuracy. We demonstrate the results of our approach using both automatic and
human evaluations. Our approach achieves some promising results for these
riddles and provides a strong baseline for future attempts. We make the entire
dataset and related materials publicly available to the community in
ImageRiddle Website (http://bit.ly/22f9Ala).
| new_dataset | 0.954393 |
1611.05915 | David Geronimo | David Ger\'onimo and Hedvig Kjellstr\"om | Generative One-Class Models for Text-based Person Retrieval in Forensic
Applications | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic forensic image analysis assists criminal investigation experts in
the search for suspicious persons, abnormal behaviors detection and identity
matching in images. In this paper we propose a person retrieval system that
uses textual queries (e.g., "black trousers and green shirt") as descriptions
and a one-class generative color model with outlier filtering to represent the
images both to train the models and to perform the search. The method is
evaluated in terms of its efficiency in fulfilling the needs of a forensic
retrieval system: limited annotation, robustness, extensibility, adaptability
and computational cost. The proposed generative method is compared to a
corresponding discriminative approach. Experiments are carried out using a
range of queries in three different databases. The experiments show that the
two evaluated algorithms provide average retrieval performance and adaptable to
new datasets. The proposed generative algorithm has some advantages over the
discriminative one, specifically its capability to work with very few training
samples and its much lower computational requirements when the number of
training examples increases.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 22:02:06 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Gerónimo",
"David",
""
],
[
"Kjellström",
"Hedvig",
""
]
] | TITLE: Generative One-Class Models for Text-based Person Retrieval in Forensic
Applications
ABSTRACT: Automatic forensic image analysis assists criminal investigation experts in
the search for suspicious persons, abnormal behaviors detection and identity
matching in images. In this paper we propose a person retrieval system that
uses textual queries (e.g., "black trousers and green shirt") as descriptions
and a one-class generative color model with outlier filtering to represent the
images both to train the models and to perform the search. The method is
evaluated in terms of its efficiency in fulfilling the needs of a forensic
retrieval system: limited annotation, robustness, extensibility, adaptability
and computational cost. The proposed generative method is compared to a
corresponding discriminative approach. Experiments are carried out using a
range of queries in three different databases. The experiments show that the
two evaluated algorithms provide average retrieval performance and adaptable to
new datasets. The proposed generative algorithm has some advantages over the
discriminative one, specifically its capability to work with very few training
samples and its much lower computational requirements when the number of
training examples increases.
| no_new_dataset | 0.950732 |
1611.06049 | Asim Ghosh Mr | Asim Ghosh, Daniel Monsivais, Kunal Bhattacharya, Robin I. M. Dunbar
and Kimmo Kaski | Quantifying gender preferences across humans lifespan | 12 pages, 8 figures | null | null | null | physics.soc-ph cs.SI q-bio.PE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In human relations individuals' gender and age play a key role in the
structures and dynamics of their social arrangements. In order to analyze the
gender preferences of individuals in interaction with others at different
stages of their lives we study a large mobile phone dataset. To do this we
consider four fundamental gender-related caller and callee combinations of
human interactions, namely male to male, male to female, female to male, and
female to female, which together with age, kinship, and different levels of
friendship give rise to a wide scope of human sociality. Here we analyse the
relative strength of these four types of interaction using a large dataset of
mobile phone communication records. Our analysis suggests strong age dependence
for an ego of one gender choosing to call an individual of either gender. We
observe a strong opposite sex bonding across most of their reproductive age.
However, older women show a strong tendency to connect to another female that
is one generation younger in a way that is suggestive of the
\emph{grandmothering effect}. We also find that the relative strength among the
four possible interactions depends on phone call duration. For calls of medium
and long duration, opposite gender interactions are significantly more probable
than same gender interactions during the reproductive years, suggesting
potential emotional exchange between spouses. By measuring the fraction of
calls to other generations we find that mothers tend to make calls more to
their daughters than to their sons, whereas fathers make calls more to their
sons than to their daughters. For younger people, most of their calls go to
same generation alters, while older people call the younger people more
frequently, which supports the suggestion that \emph{affection flows downward}.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 11:35:35 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Ghosh",
"Asim",
""
],
[
"Monsivais",
"Daniel",
""
],
[
"Bhattacharya",
"Kunal",
""
],
[
"Dunbar",
"Robin I. M.",
""
],
[
"Kaski",
"Kimmo",
""
]
] | TITLE: Quantifying gender preferences across humans lifespan
ABSTRACT: In human relations individuals' gender and age play a key role in the
structures and dynamics of their social arrangements. In order to analyze the
gender preferences of individuals in interaction with others at different
stages of their lives we study a large mobile phone dataset. To do this we
consider four fundamental gender-related caller and callee combinations of
human interactions, namely male to male, male to female, female to male, and
female to female, which together with age, kinship, and different levels of
friendship give rise to a wide scope of human sociality. Here we analyse the
relative strength of these four types of interaction using a large dataset of
mobile phone communication records. Our analysis suggests strong age dependence
for an ego of one gender choosing to call an individual of either gender. We
observe a strong opposite sex bonding across most of their reproductive age.
However, older women show a strong tendency to connect to another female that
is one generation younger in a way that is suggestive of the
\emph{grandmothering effect}. We also find that the relative strength among the
four possible interactions depends on phone call duration. For calls of medium
and long duration, opposite gender interactions are significantly more probable
than same gender interactions during the reproductive years, suggesting
potential emotional exchange between spouses. By measuring the fraction of
calls to other generations we find that mothers tend to make calls more to
their daughters than to their sons, whereas fathers make calls more to their
sons than to their daughters. For younger people, most of their calls go to
same generation alters, while older people call the younger people more
frequently, which supports the suggestion that \emph{affection flows downward}.
| no_new_dataset | 0.745537 |
1611.06067 | Sijie Song | Sijie Song, Cuiling Lan, Junliang Xing, Wenjun Zeng, Jiaying Liu | An End-to-End Spatio-Temporal Attention Model for Human Action
Recognition from Skeleton Data | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Human action recognition is an important task in computer vision. Extracting
discriminative spatial and temporal features to model the spatial and temporal
evolutions of different actions plays a key role in accomplishing this task. In
this work, we propose an end-to-end spatial and temporal attention model for
human action recognition from skeleton data. We build our model on top of the
Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM), which
learns to selectively focus on discriminative joints of skeleton within each
frame of the inputs and pays different levels of attention to the outputs of
different frames. Furthermore, to ensure effective training of the network, we
propose a regularized cross-entropy loss to drive the model learning process
and develop a joint training strategy accordingly. Experimental results
demonstrate the effectiveness of the proposed model,both on the small human
action recognition data set of SBU and the currently largest NTU dataset.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 13:33:28 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Song",
"Sijie",
""
],
[
"Lan",
"Cuiling",
""
],
[
"Xing",
"Junliang",
""
],
[
"Zeng",
"Wenjun",
""
],
[
"Liu",
"Jiaying",
""
]
] | TITLE: An End-to-End Spatio-Temporal Attention Model for Human Action
Recognition from Skeleton Data
ABSTRACT: Human action recognition is an important task in computer vision. Extracting
discriminative spatial and temporal features to model the spatial and temporal
evolutions of different actions plays a key role in accomplishing this task. In
this work, we propose an end-to-end spatial and temporal attention model for
human action recognition from skeleton data. We build our model on top of the
Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM), which
learns to selectively focus on discriminative joints of skeleton within each
frame of the inputs and pays different levels of attention to the outputs of
different frames. Furthermore, to ensure effective training of the network, we
propose a regularized cross-entropy loss to drive the model learning process
and develop a joint training strategy accordingly. Experimental results
demonstrate the effectiveness of the proposed model,both on the small human
action recognition data set of SBU and the currently largest NTU dataset.
| no_new_dataset | 0.947527 |
1611.06080 | Kian Hsiang Low | Quang Minh Hoang, Trong Nghia Hoang, Kian Hsiang Low | A Generalized Stochastic Variational Bayesian Hyperparameter Learning
Framework for Sparse Spectrum Gaussian Process Regression | 31st AAAI Conference on Artificial Intelligence (AAAI 2017), Extended
version with proofs, 11 pages | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While much research effort has been dedicated to scaling up sparse Gaussian
process (GP) models based on inducing variables for big data, little attention
is afforded to the other less explored class of low-rank GP approximations that
exploit the sparse spectral representation of a GP kernel. This paper presents
such an effort to advance the state of the art of sparse spectrum GP models to
achieve competitive predictive performance for massive datasets. Our
generalized framework of stochastic variational Bayesian sparse spectrum GP
(sVBSSGP) models addresses their shortcomings by adopting a Bayesian treatment
of the spectral frequencies to avoid overfitting, modeling these frequencies
jointly in its variational distribution to enable their interaction a
posteriori, and exploiting local data for boosting the predictive performance.
However, such structural improvements result in a variational lower bound that
is intractable to be optimized. To resolve this, we exploit a variational
parameterization trick to make it amenable to stochastic optimization.
Interestingly, the resulting stochastic gradient has a linearly decomposable
structure that can be exploited to refine our stochastic optimization method to
incur constant time per iteration while preserving its property of being an
unbiased estimator of the exact gradient of the variational lower bound.
Empirical evaluation on real-world datasets shows that sVBSSGP outperforms
state-of-the-art stochastic implementations of sparse GP models.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 14:00:48 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Hoang",
"Quang Minh",
""
],
[
"Hoang",
"Trong Nghia",
""
],
[
"Low",
"Kian Hsiang",
""
]
] | TITLE: A Generalized Stochastic Variational Bayesian Hyperparameter Learning
Framework for Sparse Spectrum Gaussian Process Regression
ABSTRACT: While much research effort has been dedicated to scaling up sparse Gaussian
process (GP) models based on inducing variables for big data, little attention
is afforded to the other less explored class of low-rank GP approximations that
exploit the sparse spectral representation of a GP kernel. This paper presents
such an effort to advance the state of the art of sparse spectrum GP models to
achieve competitive predictive performance for massive datasets. Our
generalized framework of stochastic variational Bayesian sparse spectrum GP
(sVBSSGP) models addresses their shortcomings by adopting a Bayesian treatment
of the spectral frequencies to avoid overfitting, modeling these frequencies
jointly in its variational distribution to enable their interaction a
posteriori, and exploiting local data for boosting the predictive performance.
However, such structural improvements result in a variational lower bound that
is intractable to be optimized. To resolve this, we exploit a variational
parameterization trick to make it amenable to stochastic optimization.
Interestingly, the resulting stochastic gradient has a linearly decomposable
structure that can be exploited to refine our stochastic optimization method to
incur constant time per iteration while preserving its property of being an
unbiased estimator of the exact gradient of the variational lower bound.
Empirical evaluation on real-world datasets shows that sVBSSGP outperforms
state-of-the-art stochastic implementations of sparse GP models.
| no_new_dataset | 0.946843 |
1611.06128 | Mohamed Sherif | Mohamed Ahmed Sherif, Kevin Dre{\ss}ler, Panayiotis Smeros and
Axel-Cyrille Ngonga Ngomo | Annex: Radon - Rapid Discovery of Topological Relations | 19 pages, 3 figures, i algorithm and 1 table | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Datasets containing geo-spatial resources are increasingly being represented
according to the Linked Data principles. Several time-efficient approaches for
discovering links between RDF resources have been developed over the last
years. However, the time-efficient discovery of topological relations between
geospatial resources has been paid little attention to. We address this
research gap by presenting Radon, a novel approach for the rapid computation of
topological relations between geo-spatial resources. Our approach uses a sparse
tiling index in combination with minimum bounding boxes to reduce the
computation time of topological relations. Our evaluation of Radon's runtime on
45 datasets and in more than 800 experiments shows that it outperforms the
state of the art by up to 3 orders of magnitude while maintaining an F-measure
of 100%. Moreover, our experiments suggest that Radon scales up well when
implemented in parallel.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 15:42:22 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Sherif",
"Mohamed Ahmed",
""
],
[
"Dreßler",
"Kevin",
""
],
[
"Smeros",
"Panayiotis",
""
],
[
"Ngomo",
"Axel-Cyrille Ngonga",
""
]
] | TITLE: Annex: Radon - Rapid Discovery of Topological Relations
ABSTRACT: Datasets containing geo-spatial resources are increasingly being represented
according to the Linked Data principles. Several time-efficient approaches for
discovering links between RDF resources have been developed over the last
years. However, the time-efficient discovery of topological relations between
geospatial resources has been paid little attention to. We address this
research gap by presenting Radon, a novel approach for the rapid computation of
topological relations between geo-spatial resources. Our approach uses a sparse
tiling index in combination with minimum bounding boxes to reduce the
computation time of topological relations. Our evaluation of Radon's runtime on
45 datasets and in more than 800 experiments shows that it outperforms the
state of the art by up to 3 orders of magnitude while maintaining an F-measure
of 100%. Moreover, our experiments suggest that Radon scales up well when
implemented in parallel.
| no_new_dataset | 0.946794 |
1611.06132 | Pavel Izmailov | Pavel Izmailov and Dmitry Kropotov | Faster variational inducing input Gaussian process classification | null | null | null | null | cs.LG cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Gaussian processes (GP) provide a prior over functions and allow finding
complex regularities in data. Gaussian processes are successfully used for
classification/regression problems and dimensionality reduction. In this work
we consider the classification problem only. The complexity of standard methods
for GP-classification scales cubically with the size of the training dataset.
This complexity makes them inapplicable to big data problems. Therefore, a
variety of methods were introduced to overcome this limitation. In the paper we
focus on methods based on so called inducing inputs. This approach is based on
variational inference and proposes a particular lower bound for marginal
likelihood (evidence). This bound is then maximized w.r.t. parameters of kernel
function of the Gaussian process, thus fitting the model to data. The
computational complexity of this method is $O(nm^2)$, where $m$ is the number
of inducing inputs used by the model and is assumed to be substantially smaller
than the size of the dataset $n$. Recently, a new evidence lower bound for
GP-classification problem was introduced. It allows using stochastic
optimization, which makes it suitable for big data problems. However, the new
lower bound depends on $O(m^2)$ variational parameter, which makes optimization
challenging in case of big m. In this work we develop a new approach for
training inducing input GP models for classification problems. Here we use
quadratic approximation of several terms in the aforementioned evidence lower
bound, obtaining analytical expressions for optimal values of most of the
parameters in the optimization, thus sufficiently reducing the dimension of
optimization space. In our experiments we achieve as well or better results,
compared to the existing method. Moreover, our method doesn't require the user
to manually set the learning rate, making it more practical, than the existing
method.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 15:53:50 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Izmailov",
"Pavel",
""
],
[
"Kropotov",
"Dmitry",
""
]
] | TITLE: Faster variational inducing input Gaussian process classification
ABSTRACT: Gaussian processes (GP) provide a prior over functions and allow finding
complex regularities in data. Gaussian processes are successfully used for
classification/regression problems and dimensionality reduction. In this work
we consider the classification problem only. The complexity of standard methods
for GP-classification scales cubically with the size of the training dataset.
This complexity makes them inapplicable to big data problems. Therefore, a
variety of methods were introduced to overcome this limitation. In the paper we
focus on methods based on so called inducing inputs. This approach is based on
variational inference and proposes a particular lower bound for marginal
likelihood (evidence). This bound is then maximized w.r.t. parameters of kernel
function of the Gaussian process, thus fitting the model to data. The
computational complexity of this method is $O(nm^2)$, where $m$ is the number
of inducing inputs used by the model and is assumed to be substantially smaller
than the size of the dataset $n$. Recently, a new evidence lower bound for
GP-classification problem was introduced. It allows using stochastic
optimization, which makes it suitable for big data problems. However, the new
lower bound depends on $O(m^2)$ variational parameter, which makes optimization
challenging in case of big m. In this work we develop a new approach for
training inducing input GP models for classification problems. Here we use
quadratic approximation of several terms in the aforementioned evidence lower
bound, obtaining analytical expressions for optimal values of most of the
parameters in the optimization, thus sufficiently reducing the dimension of
optimization space. In our experiments we achieve as well or better results,
compared to the existing method. Moreover, our method doesn't require the user
to manually set the learning rate, making it more practical, than the existing
method.
| no_new_dataset | 0.947672 |
1611.06211 | Mohammad Babaeizadeh | Mohammad Babaeizadeh, Paris Smaragdis, Roy H. Campbell | NoiseOut: A Simple Way to Prune Neural Networks | null | null | null | null | cs.NE cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Neural networks are usually over-parameterized with significant redundancy in
the number of required neurons which results in unnecessary computation and
memory usage at inference time. One common approach to address this issue is to
prune these big networks by removing extra neurons and parameters while
maintaining the accuracy. In this paper, we propose NoiseOut, a fully automated
pruning algorithm based on the correlation between activations of neurons in
the hidden layers. We prove that adding additional output neurons with entirely
random targets results into a higher correlation between neurons which makes
pruning by NoiseOut even more efficient. Finally, we test our method on various
networks and datasets. These experiments exhibit high pruning rates while
maintaining the accuracy of the original network.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 19:55:29 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Babaeizadeh",
"Mohammad",
""
],
[
"Smaragdis",
"Paris",
""
],
[
"Campbell",
"Roy H.",
""
]
] | TITLE: NoiseOut: A Simple Way to Prune Neural Networks
ABSTRACT: Neural networks are usually over-parameterized with significant redundancy in
the number of required neurons which results in unnecessary computation and
memory usage at inference time. One common approach to address this issue is to
prune these big networks by removing extra neurons and parameters while
maintaining the accuracy. In this paper, we propose NoiseOut, a fully automated
pruning algorithm based on the correlation between activations of neurons in
the hidden layers. We prove that adding additional output neurons with entirely
random targets results into a higher correlation between neurons which makes
pruning by NoiseOut even more efficient. Finally, we test our method on various
networks and datasets. These experiments exhibit high pruning rates while
maintaining the accuracy of the original network.
| no_new_dataset | 0.951142 |
1611.06224 | Hui Miao | Hui Miao, Ang Li, Larry S. Davis, Amol Deshpande | ModelHub: Towards Unified Data and Lifecycle Management for Deep
Learning | null | null | null | null | cs.DB cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep learning has improved state-of-the-art results in many important fields,
and has been the subject of much research in recent years, leading to the
development of several systems for facilitating deep learning. Current systems,
however, mainly focus on model building and training phases, while the issues
of data management, model sharing, and lifecycle management are largely
ignored. Deep learning modeling lifecycle generates a rich set of data
artifacts, such as learned parameters and training logs, and comprises of
several frequently conducted tasks, e.g., to understand the model behaviors and
to try out new models. Dealing with such artifacts and tasks is cumbersome and
largely left to the users. This paper describes our vision and implementation
of a data and lifecycle management system for deep learning. First, we
generalize model exploration and model enumeration queries from commonly
conducted tasks by deep learning modelers, and propose a high-level domain
specific language (DSL), inspired by SQL, to raise the abstraction level and
accelerate the modeling process. To manage the data artifacts, especially the
large amount of checkpointed float parameters, we design a novel model
versioning system (dlv), and a read-optimized parameter archival storage system
(PAS) that minimizes storage footprint and accelerates query workloads without
losing accuracy. PAS archives versioned models using deltas in a
multi-resolution fashion by separately storing the less significant bits, and
features a novel progressive query (inference) evaluation algorithm. Third, we
show that archiving versioned models using deltas poses a new dataset
versioning problem and we develop efficient algorithms for solving it. We
conduct extensive experiments over several real datasets from computer vision
domain to show the efficiency of the proposed techniques.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 20:59:25 GMT"
}
] | 2016-11-21T00:00:00 | [
[
"Miao",
"Hui",
""
],
[
"Li",
"Ang",
""
],
[
"Davis",
"Larry S.",
""
],
[
"Deshpande",
"Amol",
""
]
] | TITLE: ModelHub: Towards Unified Data and Lifecycle Management for Deep
Learning
ABSTRACT: Deep learning has improved state-of-the-art results in many important fields,
and has been the subject of much research in recent years, leading to the
development of several systems for facilitating deep learning. Current systems,
however, mainly focus on model building and training phases, while the issues
of data management, model sharing, and lifecycle management are largely
ignored. Deep learning modeling lifecycle generates a rich set of data
artifacts, such as learned parameters and training logs, and comprises of
several frequently conducted tasks, e.g., to understand the model behaviors and
to try out new models. Dealing with such artifacts and tasks is cumbersome and
largely left to the users. This paper describes our vision and implementation
of a data and lifecycle management system for deep learning. First, we
generalize model exploration and model enumeration queries from commonly
conducted tasks by deep learning modelers, and propose a high-level domain
specific language (DSL), inspired by SQL, to raise the abstraction level and
accelerate the modeling process. To manage the data artifacts, especially the
large amount of checkpointed float parameters, we design a novel model
versioning system (dlv), and a read-optimized parameter archival storage system
(PAS) that minimizes storage footprint and accelerates query workloads without
losing accuracy. PAS archives versioned models using deltas in a
multi-resolution fashion by separately storing the less significant bits, and
features a novel progressive query (inference) evaluation algorithm. Third, we
show that archiving versioned models using deltas poses a new dataset
versioning problem and we develop efficient algorithms for solving it. We
conduct extensive experiments over several real datasets from computer vision
domain to show the efficiency of the proposed techniques.
| no_new_dataset | 0.952794 |
1006.1772 | Kishor Barman | Kishor Barman, Onkar Dabeer | Analysis of a Collaborative Filter Based on Popularity Amongst Neighbors | 47 pages. Submitted to IEEE Transactions on Information Theory
(revised in July 2011). A shorter version would be presented at ISIT 2010 | null | 10.1109/TIT.2012.2216980 | null | cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we analyze a collaborative filter that answers the simple
question: What is popular amongst your friends? While this basic principle
seems to be prevalent in many practical implementations, there does not appear
to be much theoretical analysis of its performance. In this paper, we partly
fill this gap. While recent works on this topic, such as the low-rank matrix
completion literature, consider the probability of error in recovering the
entire rating matrix, we consider probability of an error in an individual
recommendation (bit error rate (BER)). For a mathematical model introduced in
[1],[2], we identify three regimes of operation for our algorithm (named
Popularity Amongst Friends (PAF)) in the limit as the matrix size grows to
infinity. In a regime characterized by large number of samples and small
degrees of freedom (defined precisely for the model in the paper), the
asymptotic BER is zero; in a regime characterized by large number of samples
and large degrees of freedom, the asymptotic BER is bounded away from 0 and 1/2
(and is identified exactly except for a special case); and in a regime
characterized by a small number of samples, the algorithm fails. We also
present numerical results for the MovieLens and Netflix datasets. We discuss
the empirical performance in light of our theoretical results and compare with
an approach based on low-rank matrix completion.
| [
{
"version": "v1",
"created": "Wed, 9 Jun 2010 11:48:53 GMT"
},
{
"version": "v2",
"created": "Thu, 14 Jul 2011 10:58:41 GMT"
},
{
"version": "v3",
"created": "Fri, 15 Jul 2011 07:29:19 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Barman",
"Kishor",
""
],
[
"Dabeer",
"Onkar",
""
]
] | TITLE: Analysis of a Collaborative Filter Based on Popularity Amongst Neighbors
ABSTRACT: In this paper, we analyze a collaborative filter that answers the simple
question: What is popular amongst your friends? While this basic principle
seems to be prevalent in many practical implementations, there does not appear
to be much theoretical analysis of its performance. In this paper, we partly
fill this gap. While recent works on this topic, such as the low-rank matrix
completion literature, consider the probability of error in recovering the
entire rating matrix, we consider probability of an error in an individual
recommendation (bit error rate (BER)). For a mathematical model introduced in
[1],[2], we identify three regimes of operation for our algorithm (named
Popularity Amongst Friends (PAF)) in the limit as the matrix size grows to
infinity. In a regime characterized by large number of samples and small
degrees of freedom (defined precisely for the model in the paper), the
asymptotic BER is zero; in a regime characterized by large number of samples
and large degrees of freedom, the asymptotic BER is bounded away from 0 and 1/2
(and is identified exactly except for a special case); and in a regime
characterized by a small number of samples, the algorithm fails. We also
present numerical results for the MovieLens and Netflix datasets. We discuss
the empirical performance in light of our theoretical results and compare with
an approach based on low-rank matrix completion.
| no_new_dataset | 0.949669 |
1009.2275 | Anh Le | Anh Le, Athina Markopoulou, Michalis Faloutsos | PhishDef: URL Names Say It All | 9 pages, submitted to IEEE INFOCOM 2011 | null | 10.1109/INFCOM.2011.5934995 | null | cs.CR cs.LG cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Phishing is an increasingly sophisticated method to steal personal user
information using sites that pretend to be legitimate. In this paper, we take
the following steps to identify phishing URLs. First, we carefully select
lexical features of the URLs that are resistant to obfuscation techniques used
by attackers. Second, we evaluate the classification accuracy when using only
lexical features, both automatically and hand-selected, vs. when using
additional features. We show that lexical features are sufficient for all
practical purposes. Third, we thoroughly compare several classification
algorithms, and we propose to use an online method (AROW) that is able to
overcome noisy training data. Based on the insights gained from our analysis,
we propose PhishDef, a phishing detection system that uses only URL names and
combines the above three elements. PhishDef is a highly accurate method (when
compared to state-of-the-art approaches over real datasets), lightweight (thus
appropriate for online and client-side deployment), proactive (based on online
classification rather than blacklists), and resilient to training data
inaccuracies (thus enabling the use of large noisy training data).
| [
{
"version": "v1",
"created": "Sun, 12 Sep 2010 23:55:00 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Le",
"Anh",
""
],
[
"Markopoulou",
"Athina",
""
],
[
"Faloutsos",
"Michalis",
""
]
] | TITLE: PhishDef: URL Names Say It All
ABSTRACT: Phishing is an increasingly sophisticated method to steal personal user
information using sites that pretend to be legitimate. In this paper, we take
the following steps to identify phishing URLs. First, we carefully select
lexical features of the URLs that are resistant to obfuscation techniques used
by attackers. Second, we evaluate the classification accuracy when using only
lexical features, both automatically and hand-selected, vs. when using
additional features. We show that lexical features are sufficient for all
practical purposes. Third, we thoroughly compare several classification
algorithms, and we propose to use an online method (AROW) that is able to
overcome noisy training data. Based on the insights gained from our analysis,
we propose PhishDef, a phishing detection system that uses only URL names and
combines the above three elements. PhishDef is a highly accurate method (when
compared to state-of-the-art approaches over real datasets), lightweight (thus
appropriate for online and client-side deployment), proactive (based on online
classification rather than blacklists), and resilient to training data
inaccuracies (thus enabling the use of large noisy training data).
| no_new_dataset | 0.948202 |
1012.3189 | Jian Li | Jian Li, Amol Deshpande | Maximizing Expected Utility for Stochastic Combinatorial Optimization
Problems | 31 pages, Preliminary version appears in the Proceeding of the 52nd
Annual IEEE Symposium on Foundations of Computer Science (FOCS 2011), This
version contains several new results ( results (2) and (3) in the abstract) | null | 10.1109/FOCS.2011.33 | null | cs.DS cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the stochastic versions of a broad class of combinatorial problems
where the weights of the elements in the input dataset are uncertain. The class
of problems that we study includes shortest paths, minimum weight spanning
trees, and minimum weight matchings, and other combinatorial problems like
knapsack. We observe that the expected value is inadequate in capturing
different types of {\em risk-averse} or {\em risk-prone} behaviors, and instead
we consider a more general objective which is to maximize the {\em expected
utility} of the solution for some given utility function, rather than the
expected weight (expected weight becomes a special case). Under the assumption
that there is a pseudopolynomial time algorithm for the {\em exact} version of
the problem (This is true for the problems mentioned above), we can obtain the
following approximation results for several important classes of utility
functions: (1) If the utility function $\uti$ is continuous, upper-bounded by a
constant and $\lim_{x\rightarrow+\infty}\uti(x)=0$, we show that we can obtain
a polynomial time approximation algorithm with an {\em additive error}
$\epsilon$ for any constant $\epsilon>0$. (2) If the utility function $\uti$ is
a concave increasing function, we can obtain a polynomial time approximation
scheme (PTAS). (3) If the utility function $\uti$ is increasing and has a
bounded derivative, we can obtain a polynomial time approximation scheme. Our
results recover or generalize several prior results on stochastic shortest
path, stochastic spanning tree, and stochastic knapsack. Our algorithm for
utility maximization makes use of the separability of exponential utility and a
technique to decompose a general utility function into exponential utility
functions, which may be useful in other stochastic optimization problems.
| [
{
"version": "v1",
"created": "Tue, 14 Dec 2010 22:34:32 GMT"
},
{
"version": "v2",
"created": "Fri, 15 Apr 2011 01:10:47 GMT"
},
{
"version": "v3",
"created": "Sun, 14 Aug 2011 02:53:22 GMT"
},
{
"version": "v4",
"created": "Fri, 19 Aug 2011 10:32:48 GMT"
},
{
"version": "v5",
"created": "Tue, 19 Mar 2013 09:11:30 GMT"
},
{
"version": "v6",
"created": "Sat, 19 Dec 2015 10:13:56 GMT"
},
{
"version": "v7",
"created": "Wed, 10 Aug 2016 09:02:50 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Li",
"Jian",
""
],
[
"Deshpande",
"Amol",
""
]
] | TITLE: Maximizing Expected Utility for Stochastic Combinatorial Optimization
Problems
ABSTRACT: We study the stochastic versions of a broad class of combinatorial problems
where the weights of the elements in the input dataset are uncertain. The class
of problems that we study includes shortest paths, minimum weight spanning
trees, and minimum weight matchings, and other combinatorial problems like
knapsack. We observe that the expected value is inadequate in capturing
different types of {\em risk-averse} or {\em risk-prone} behaviors, and instead
we consider a more general objective which is to maximize the {\em expected
utility} of the solution for some given utility function, rather than the
expected weight (expected weight becomes a special case). Under the assumption
that there is a pseudopolynomial time algorithm for the {\em exact} version of
the problem (This is true for the problems mentioned above), we can obtain the
following approximation results for several important classes of utility
functions: (1) If the utility function $\uti$ is continuous, upper-bounded by a
constant and $\lim_{x\rightarrow+\infty}\uti(x)=0$, we show that we can obtain
a polynomial time approximation algorithm with an {\em additive error}
$\epsilon$ for any constant $\epsilon>0$. (2) If the utility function $\uti$ is
a concave increasing function, we can obtain a polynomial time approximation
scheme (PTAS). (3) If the utility function $\uti$ is increasing and has a
bounded derivative, we can obtain a polynomial time approximation scheme. Our
results recover or generalize several prior results on stochastic shortest
path, stochastic spanning tree, and stochastic knapsack. Our algorithm for
utility maximization makes use of the separability of exponential utility and a
technique to decompose a general utility function into exponential utility
functions, which may be useful in other stochastic optimization problems.
| no_new_dataset | 0.948822 |
1105.2264 | Artem Chebotko | Craig Franke, Samuel Morin, Artem Chebotko, John Abraham, Pearl
Brazier | Distributed Semantic Web Data Management in HBase and MySQL Cluster | In Proc. of the 4th IEEE International Conference on Cloud Computing
(CLOUD'11) | null | 10.1109/CLOUD.2011.19 | null | cs.DB cs.PF | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Various computing and data resources on the Web are being enhanced with
machine-interpretable semantic descriptions to facilitate better search,
discovery and integration. This interconnected metadata constitutes the
Semantic Web, whose volume can potentially grow the scale of the Web. Efficient
management of Semantic Web data, expressed using the W3C's Resource Description
Framework (RDF), is crucial for supporting new data-intensive,
semantics-enabled applications. In this work, we study and compare two
approaches to distributed RDF data management based on emerging cloud computing
technologies and traditional relational database clustering technologies. In
particular, we design distributed RDF data storage and querying schemes for
HBase and MySQL Cluster and conduct an empirical comparison of these approaches
on a cluster of commodity machines using datasets and queries from the Third
Provenance Challenge and Lehigh University Benchmark. Our study reveals
interesting patterns in query evaluation, shows that our algorithms are
promising, and suggests that cloud computing has a great potential for scalable
Semantic Web data management.
| [
{
"version": "v1",
"created": "Wed, 11 May 2011 17:46:15 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Franke",
"Craig",
""
],
[
"Morin",
"Samuel",
""
],
[
"Chebotko",
"Artem",
""
],
[
"Abraham",
"John",
""
],
[
"Brazier",
"Pearl",
""
]
] | TITLE: Distributed Semantic Web Data Management in HBase and MySQL Cluster
ABSTRACT: Various computing and data resources on the Web are being enhanced with
machine-interpretable semantic descriptions to facilitate better search,
discovery and integration. This interconnected metadata constitutes the
Semantic Web, whose volume can potentially grow the scale of the Web. Efficient
management of Semantic Web data, expressed using the W3C's Resource Description
Framework (RDF), is crucial for supporting new data-intensive,
semantics-enabled applications. In this work, we study and compare two
approaches to distributed RDF data management based on emerging cloud computing
technologies and traditional relational database clustering technologies. In
particular, we design distributed RDF data storage and querying schemes for
HBase and MySQL Cluster and conduct an empirical comparison of these approaches
on a cluster of commodity machines using datasets and queries from the Third
Provenance Challenge and Lehigh University Benchmark. Our study reveals
interesting patterns in query evaluation, shows that our algorithms are
promising, and suggests that cloud computing has a great potential for scalable
Semantic Web data management.
| no_new_dataset | 0.947914 |
1207.3285 | Sarvesh Nikumbh | Sarvesh Nikumbh, Shameek Ghosh, Valadi Jayaraman | Biogeography-Based Informative Gene Selection and Cancer Classification
Using SVM and Random Forests | 6 pages; Author's copy; Presented at the IEEE World Congress on
Computational Intelligence (at IEEE Congress on Evolutionary Computation),
Brisbane, Australia, June 2012 | null | 10.1109/CEC.2012.6256127 | CMS-TR-20120509 | cs.NE stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Microarray cancer gene expression data comprise of very high dimensions.
Reducing the dimensions helps in improving the overall analysis and
classification performance. We propose two hybrid techniques, Biogeography -
based Optimization - Random Forests (BBO - RF) and BBO - SVM (Support Vector
Machines) with gene ranking as a heuristic, for microarray gene expression
analysis. This heuristic is obtained from information gain filter ranking
procedure. The BBO algorithm generates a population of candidate subset of
genes, as part of an ecosystem of habitats, and employs the migration and
mutation processes across multiple generations of the population to improve the
classification accuracy. The fitness of each gene subset is assessed by the
classifiers - SVM and Random Forests. The performances of these hybrid
techniques are evaluated on three cancer gene expression datasets retrieved
from the Kent Ridge Biomedical datasets collection and the libSVM data
repository. Our results demonstrate that genes selected by the proposed
techniques yield classification accuracies comparable to previously reported
algorithms.
| [
{
"version": "v1",
"created": "Thu, 12 Jul 2012 19:00:46 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Nikumbh",
"Sarvesh",
""
],
[
"Ghosh",
"Shameek",
""
],
[
"Jayaraman",
"Valadi",
""
]
] | TITLE: Biogeography-Based Informative Gene Selection and Cancer Classification
Using SVM and Random Forests
ABSTRACT: Microarray cancer gene expression data comprise of very high dimensions.
Reducing the dimensions helps in improving the overall analysis and
classification performance. We propose two hybrid techniques, Biogeography -
based Optimization - Random Forests (BBO - RF) and BBO - SVM (Support Vector
Machines) with gene ranking as a heuristic, for microarray gene expression
analysis. This heuristic is obtained from information gain filter ranking
procedure. The BBO algorithm generates a population of candidate subset of
genes, as part of an ecosystem of habitats, and employs the migration and
mutation processes across multiple generations of the population to improve the
classification accuracy. The fitness of each gene subset is assessed by the
classifiers - SVM and Random Forests. The performances of these hybrid
techniques are evaluated on three cancer gene expression datasets retrieved
from the Kent Ridge Biomedical datasets collection and the libSVM data
repository. Our results demonstrate that genes selected by the proposed
techniques yield classification accuracies comparable to previously reported
algorithms.
| no_new_dataset | 0.955693 |
1210.4211 | Wei Lu | Wei Lu, Laks V.S. Lakshmanan | Profit Maximization over Social Networks | 19 pages, 8 figures. An abbreviated version appears in 2012 IEEE
International Conference on Data Mining (ICDM'12). The second version
includes some minor fixes | null | 10.1109/ICDM.2012.145 | null | cs.SI cs.GT physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Influence maximization is the problem of finding a set of influential users
in a social network such that the expected spread of influence under a certain
propagation model is maximized. Much of the previous work has neglected the
important distinction between social influence and actual product adoption.
However, as recognized in the management science literature, an individual who
gets influenced by social acquaintances may not necessarily adopt a product (or
technology), due, e.g., to monetary concerns. In this work, we distinguish
between influence and adoption by explicitly modeling the states of being
influenced and of adopting a product. We extend the classical Linear Threshold
(LT) model to incorporate prices and valuations, and factor them into users'
decision-making process of adopting a product. We show that the expected profit
function under our proposed model maintains submodularity under certain
conditions, but no longer exhibits monotonicity, unlike the expected influence
spread function. To maximize the expected profit under our extended LT model,
we employ an unbudgeted greedy framework to propose three profit maximization
algorithms. The results of our detailed experimental study on three real-world
datasets demonstrate that of the three algorithms, \textsf{PAGE}, which assigns
prices dynamically based on the profit potential of each candidate seed, has
the best performance both in the expected profit achieved and in running time.
| [
{
"version": "v1",
"created": "Mon, 15 Oct 2012 22:32:37 GMT"
},
{
"version": "v2",
"created": "Wed, 5 Jun 2013 16:41:04 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Lu",
"Wei",
""
],
[
"Lakshmanan",
"Laks V. S.",
""
]
] | TITLE: Profit Maximization over Social Networks
ABSTRACT: Influence maximization is the problem of finding a set of influential users
in a social network such that the expected spread of influence under a certain
propagation model is maximized. Much of the previous work has neglected the
important distinction between social influence and actual product adoption.
However, as recognized in the management science literature, an individual who
gets influenced by social acquaintances may not necessarily adopt a product (or
technology), due, e.g., to monetary concerns. In this work, we distinguish
between influence and adoption by explicitly modeling the states of being
influenced and of adopting a product. We extend the classical Linear Threshold
(LT) model to incorporate prices and valuations, and factor them into users'
decision-making process of adopting a product. We show that the expected profit
function under our proposed model maintains submodularity under certain
conditions, but no longer exhibits monotonicity, unlike the expected influence
spread function. To maximize the expected profit under our extended LT model,
we employ an unbudgeted greedy framework to propose three profit maximization
algorithms. The results of our detailed experimental study on three real-world
datasets demonstrate that of the three algorithms, \textsf{PAGE}, which assigns
prices dynamically based on the profit potential of each candidate seed, has
the best performance both in the expected profit achieved and in running time.
| no_new_dataset | 0.947235 |
1308.2565 | Chlo\"e Brown | Chlo\"e Brown, Anastasios Noulas, Cecilia Mascolo, Vincent Blondel | A place-focused model for social networks in cities | 13 pages, 7 figures. IEEE/ASE SocialCom 2013 | null | 10.1109/SocialCom.2013.18 | null | cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The focused organization theory of social ties proposes that the structure of
human social networks can be arranged around extra-network foci, which can
include shared physical spaces such as homes, workplaces, restaurants, and so
on. Until now, this has been difficult to investigate on a large scale, but the
huge volume of data available from online location-based social services now
makes it possible to examine the friendships and mobility of many thousands of
people, and to investigate the relationship between meetings at places and the
structure of the social network. In this paper, we analyze a large dataset from
Foursquare, the most popular online location-based social network. We examine
the properties of city-based social networks, finding that they have common
structural properties, and that the category of place where two people meet has
very strong influence on the likelihood of their being friends. Inspired by
these observations in combination with the focused organization theory, we then
present a model to generate city-level social networks, and show that it
produces networks with the structural properties seen in empirical data.
| [
{
"version": "v1",
"created": "Mon, 12 Aug 2013 13:58:15 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Brown",
"Chloë",
""
],
[
"Noulas",
"Anastasios",
""
],
[
"Mascolo",
"Cecilia",
""
],
[
"Blondel",
"Vincent",
""
]
] | TITLE: A place-focused model for social networks in cities
ABSTRACT: The focused organization theory of social ties proposes that the structure of
human social networks can be arranged around extra-network foci, which can
include shared physical spaces such as homes, workplaces, restaurants, and so
on. Until now, this has been difficult to investigate on a large scale, but the
huge volume of data available from online location-based social services now
makes it possible to examine the friendships and mobility of many thousands of
people, and to investigate the relationship between meetings at places and the
structure of the social network. In this paper, we analyze a large dataset from
Foursquare, the most popular online location-based social network. We examine
the properties of city-based social networks, finding that they have common
structural properties, and that the category of place where two people meet has
very strong influence on the likelihood of their being friends. Inspired by
these observations in combination with the focused organization theory, we then
present a model to generate city-level social networks, and show that it
produces networks with the structural properties seen in empirical data.
| no_new_dataset | 0.953966 |
1309.3908 | Marco Guerini | Marco Guerini, Jacopo Staiano, Davide Albanese | Exploring Image Virality in Google Plus | 8 pages, 8 figures. IEEE/ASE SocialCom 2013 | null | 10.1109/SocialCom.2013.101 | null | cs.SI cs.CY cs.MM physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reactions to posts in an online social network show different dynamics
depending on several textual features of the corresponding content. Do similar
dynamics exist when images are posted? Exploiting a novel dataset of posts,
gathered from the most popular Google+ users, we try to give an answer to such
a question. We describe several virality phenomena that emerge when taking into
account visual characteristics of images (such as orientation, mean saturation,
etc.). We also provide hypotheses and potential explanations for the dynamics
behind them, and include cases for which common-sense expectations do not hold
true in our experiments.
| [
{
"version": "v1",
"created": "Mon, 16 Sep 2013 11:35:19 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Guerini",
"Marco",
""
],
[
"Staiano",
"Jacopo",
""
],
[
"Albanese",
"Davide",
""
]
] | TITLE: Exploring Image Virality in Google Plus
ABSTRACT: Reactions to posts in an online social network show different dynamics
depending on several textual features of the corresponding content. Do similar
dynamics exist when images are posted? Exploiting a novel dataset of posts,
gathered from the most popular Google+ users, we try to give an answer to such
a question. We describe several virality phenomena that emerge when taking into
account visual characteristics of images (such as orientation, mean saturation,
etc.). We also provide hypotheses and potential explanations for the dynamics
behind them, and include cases for which common-sense expectations do not hold
true in our experiments.
| new_dataset | 0.951863 |
1402.2699 | Neil Zhenqiang Gong | Neil Zhenqiang Gong, Di Wang | On the Security of Trustee-based Social Authentications | 13 pages. Accepted by IEEE Transactions on Information Forensics and
Security | null | 10.1109/TIFS.2014.2330311 | null | cs.CR cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, authenticating users with the help of their friends (i.e.,
trustee-based social authentication) has been shown to be a promising backup
authentication mechanism. A user in this system is associated with a few
trustees that were selected from the user's friends. When the user wants to
regain access to the account, the service provider sends different verification
codes to the user's trustees. The user must obtain at least k (i.e., recovery
threshold) verification codes from the trustees before being directed to reset
his or her password.
In this paper, we provide the first systematic study about the security of
trustee-based social authentications. Specifically, we first introduce a novel
framework of attacks, which we call forest fire attacks. In these attacks, an
attacker initially obtains a small number of compromised users, and then the
attacker iteratively attacks the rest of users by exploiting trustee-based
social authentications. Then, we construct a probabilistic model to formalize
the threats of forest fire attacks and their costs for attackers. Moreover, we
introduce various defense strategies. Finally, we apply our framework to
extensively evaluate various concrete attack and defense strategies using three
real-world social network datasets. Our results have strong implications for
the design of more secure trustee-based social authentications.
| [
{
"version": "v1",
"created": "Wed, 12 Feb 2014 00:10:59 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Feb 2014 23:31:41 GMT"
},
{
"version": "v3",
"created": "Sun, 16 Mar 2014 02:56:04 GMT"
},
{
"version": "v4",
"created": "Sun, 8 Jun 2014 00:26:55 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Gong",
"Neil Zhenqiang",
""
],
[
"Wang",
"Di",
""
]
] | TITLE: On the Security of Trustee-based Social Authentications
ABSTRACT: Recently, authenticating users with the help of their friends (i.e.,
trustee-based social authentication) has been shown to be a promising backup
authentication mechanism. A user in this system is associated with a few
trustees that were selected from the user's friends. When the user wants to
regain access to the account, the service provider sends different verification
codes to the user's trustees. The user must obtain at least k (i.e., recovery
threshold) verification codes from the trustees before being directed to reset
his or her password.
In this paper, we provide the first systematic study about the security of
trustee-based social authentications. Specifically, we first introduce a novel
framework of attacks, which we call forest fire attacks. In these attacks, an
attacker initially obtains a small number of compromised users, and then the
attacker iteratively attacks the rest of users by exploiting trustee-based
social authentications. Then, we construct a probabilistic model to formalize
the threats of forest fire attacks and their costs for attackers. Moreover, we
introduce various defense strategies. Finally, we apply our framework to
extensively evaluate various concrete attack and defense strategies using three
real-world social network datasets. Our results have strong implications for
the design of more secure trustee-based social authentications.
| no_new_dataset | 0.939471 |
1406.4729 | Kaiming He | Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun | Spatial Pyramid Pooling in Deep Convolutional Networks for Visual
Recognition | This manuscript is the accepted version for IEEE Transactions on
Pattern Analysis and Machine Intelligence (TPAMI) 2015. See Changelog | null | 10.1007/978-3-319-10578-9_23 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Existing deep convolutional neural networks (CNNs) require a fixed-size
(e.g., 224x224) input image. This requirement is "artificial" and may reduce
the recognition accuracy for the images or sub-images of an arbitrary
size/scale. In this work, we equip the networks with another pooling strategy,
"spatial pyramid pooling", to eliminate the above requirement. The new network
structure, called SPP-net, can generate a fixed-length representation
regardless of image size/scale. Pyramid pooling is also robust to object
deformations. With these advantages, SPP-net should in general improve all
CNN-based image classification methods. On the ImageNet 2012 dataset, we
demonstrate that SPP-net boosts the accuracy of a variety of CNN architectures
despite their different designs. On the Pascal VOC 2007 and Caltech101
datasets, SPP-net achieves state-of-the-art classification results using a
single full-image representation and no fine-tuning.
The power of SPP-net is also significant in object detection. Using SPP-net,
we compute the feature maps from the entire image only once, and then pool
features in arbitrary regions (sub-images) to generate fixed-length
representations for training the detectors. This method avoids repeatedly
computing the convolutional features. In processing test images, our method is
24-102x faster than the R-CNN method, while achieving better or comparable
accuracy on Pascal VOC 2007.
In ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2014, our
methods rank #2 in object detection and #3 in image classification among all 38
teams. This manuscript also introduces the improvement made for this
competition.
| [
{
"version": "v1",
"created": "Wed, 18 Jun 2014 14:24:17 GMT"
},
{
"version": "v2",
"created": "Fri, 29 Aug 2014 10:28:55 GMT"
},
{
"version": "v3",
"created": "Tue, 6 Jan 2015 07:16:54 GMT"
},
{
"version": "v4",
"created": "Thu, 23 Apr 2015 07:33:24 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"He",
"Kaiming",
""
],
[
"Zhang",
"Xiangyu",
""
],
[
"Ren",
"Shaoqing",
""
],
[
"Sun",
"Jian",
""
]
] | TITLE: Spatial Pyramid Pooling in Deep Convolutional Networks for Visual
Recognition
ABSTRACT: Existing deep convolutional neural networks (CNNs) require a fixed-size
(e.g., 224x224) input image. This requirement is "artificial" and may reduce
the recognition accuracy for the images or sub-images of an arbitrary
size/scale. In this work, we equip the networks with another pooling strategy,
"spatial pyramid pooling", to eliminate the above requirement. The new network
structure, called SPP-net, can generate a fixed-length representation
regardless of image size/scale. Pyramid pooling is also robust to object
deformations. With these advantages, SPP-net should in general improve all
CNN-based image classification methods. On the ImageNet 2012 dataset, we
demonstrate that SPP-net boosts the accuracy of a variety of CNN architectures
despite their different designs. On the Pascal VOC 2007 and Caltech101
datasets, SPP-net achieves state-of-the-art classification results using a
single full-image representation and no fine-tuning.
The power of SPP-net is also significant in object detection. Using SPP-net,
we compute the feature maps from the entire image only once, and then pool
features in arbitrary regions (sub-images) to generate fixed-length
representations for training the detectors. This method avoids repeatedly
computing the convolutional features. In processing test images, our method is
24-102x faster than the R-CNN method, while achieving better or comparable
accuracy on Pascal VOC 2007.
In ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2014, our
methods rank #2 in object detection and #3 in image classification among all 38
teams. This manuscript also introduces the improvement made for this
competition.
| no_new_dataset | 0.949716 |
1410.2707 | Giovanni Caudullo | Giovanni Caudullo | Applying Geospatial Semantic Array Programming for a Reproducible Set of
Bioclimatic Indices in Europe | 10 pages, 4 figures, 1 table, published in IEEE Earthzine 2014 Vol. 7
Issue 2, 877975+ 2nd quarter theme. Geospatial Semantic Array Programming.
Available: http://www.earthzine.org/?p=877975 | IEEE Earthzine, vol. 7, no. 2, pp. 877 975+, 2014 | 10.1101/009589 | null | cs.CE physics.ao-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Bioclimate-driven regression analysis is a widely used approach for modelling
ecological niches and zonation. Although the bioclimatic complexity of the
European continent is high, a particular combination of 12 climatic and
topographic covariates was recently found able to reliably reproduce the
ecological zoning of the Food and Agriculture Organization of the United
Nations (FAO) for forest resources assessment at pan-European scale, generating
the first fuzzy similarity map of FAO ecozones in Europe. The reproducible
procedure followed to derive this collection of bioclimatic indices is now
presented. It required an integration of data-transformation modules (D-TM)
using geospatial tools such as Geographic Information System (GIS) software,
and array-based mathematical implementation such as semantic array programming
(SemAP). Base variables, intermediate and final covariates are described and
semantically defined by providing the workflow of D-TMs and the mathematical
formulation following the SemAP notation. Source layers to derive base
variables were extracted by exclusively relying on global-scale public open
geodata in order for the same set of bioclimatic covariates to be reproducible
in any region worldwide. In particular, two freely available datasets were
exploited for temperature and precipitation (WorldClim) and elevation (Global
Multi-resolution Terrain Elevation Data). The working extent covers the
European continent to the Urals with a resolution of 30 arc-second. The
proposed set of bioclimatic covariates will be made available as open data in
the European Forest Data Centre (EFDAC). The forthcoming complete set of D-TM
codelets will enable the 12 covariates to be easily reproduced and expanded
through free software.
| [
{
"version": "v1",
"created": "Fri, 10 Oct 2014 08:20:10 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Caudullo",
"Giovanni",
""
]
] | TITLE: Applying Geospatial Semantic Array Programming for a Reproducible Set of
Bioclimatic Indices in Europe
ABSTRACT: Bioclimate-driven regression analysis is a widely used approach for modelling
ecological niches and zonation. Although the bioclimatic complexity of the
European continent is high, a particular combination of 12 climatic and
topographic covariates was recently found able to reliably reproduce the
ecological zoning of the Food and Agriculture Organization of the United
Nations (FAO) for forest resources assessment at pan-European scale, generating
the first fuzzy similarity map of FAO ecozones in Europe. The reproducible
procedure followed to derive this collection of bioclimatic indices is now
presented. It required an integration of data-transformation modules (D-TM)
using geospatial tools such as Geographic Information System (GIS) software,
and array-based mathematical implementation such as semantic array programming
(SemAP). Base variables, intermediate and final covariates are described and
semantically defined by providing the workflow of D-TMs and the mathematical
formulation following the SemAP notation. Source layers to derive base
variables were extracted by exclusively relying on global-scale public open
geodata in order for the same set of bioclimatic covariates to be reproducible
in any region worldwide. In particular, two freely available datasets were
exploited for temperature and precipitation (WorldClim) and elevation (Global
Multi-resolution Terrain Elevation Data). The working extent covers the
European continent to the Urals with a resolution of 30 arc-second. The
proposed set of bioclimatic covariates will be made available as open data in
the European Forest Data Centre (EFDAC). The forthcoming complete set of D-TM
codelets will enable the 12 covariates to be easily reproduced and expanded
through free software.
| no_new_dataset | 0.956715 |
1502.06105 | Taehoon Lee | Taehoon Lee, Taesup Moon, Seung Jean Kim, Sungroh Yoon | Regularization and Kernelization of the Maximin Correlation Approach | Submitted to IEEE Access | null | 10.1109/ACCESS.2016.2551727 | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Robust classification becomes challenging when each class consists of
multiple subclasses. Examples include multi-font optical character recognition
and automated protein function prediction. In correlation-based
nearest-neighbor classification, the maximin correlation approach (MCA)
provides the worst-case optimal solution by minimizing the maximum
misclassification risk through an iterative procedure. Despite the optimality,
the original MCA has drawbacks that have limited its wide applicability in
practice. That is, the MCA tends to be sensitive to outliers, cannot
effectively handle nonlinearities in datasets, and suffers from having high
computational complexity. To address these limitations, we propose an improved
solution, named regularized maximin correlation approach (R-MCA). We first
reformulate MCA as a quadratically constrained linear programming (QCLP)
problem, incorporate regularization by introducing slack variables in the
primal problem of the QCLP, and derive the corresponding Lagrangian dual. The
dual formulation enables us to apply the kernel trick to R-MCA so that it can
better handle nonlinearities. Our experimental results demonstrate that the
regularization and kernelization make the proposed R-MCA more robust and
accurate for various classification tasks than the original MCA. Furthermore,
when the data size or dimensionality grows, R-MCA runs substantially faster by
solving either the primal or dual (whichever has a smaller variable dimension)
of the QCLP.
| [
{
"version": "v1",
"created": "Sat, 21 Feb 2015 14:37:44 GMT"
},
{
"version": "v2",
"created": "Tue, 29 Mar 2016 04:42:12 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Lee",
"Taehoon",
""
],
[
"Moon",
"Taesup",
""
],
[
"Kim",
"Seung Jean",
""
],
[
"Yoon",
"Sungroh",
""
]
] | TITLE: Regularization and Kernelization of the Maximin Correlation Approach
ABSTRACT: Robust classification becomes challenging when each class consists of
multiple subclasses. Examples include multi-font optical character recognition
and automated protein function prediction. In correlation-based
nearest-neighbor classification, the maximin correlation approach (MCA)
provides the worst-case optimal solution by minimizing the maximum
misclassification risk through an iterative procedure. Despite the optimality,
the original MCA has drawbacks that have limited its wide applicability in
practice. That is, the MCA tends to be sensitive to outliers, cannot
effectively handle nonlinearities in datasets, and suffers from having high
computational complexity. To address these limitations, we propose an improved
solution, named regularized maximin correlation approach (R-MCA). We first
reformulate MCA as a quadratically constrained linear programming (QCLP)
problem, incorporate regularization by introducing slack variables in the
primal problem of the QCLP, and derive the corresponding Lagrangian dual. The
dual formulation enables us to apply the kernel trick to R-MCA so that it can
better handle nonlinearities. Our experimental results demonstrate that the
regularization and kernelization make the proposed R-MCA more robust and
accurate for various classification tasks than the original MCA. Furthermore,
when the data size or dimensionality grows, R-MCA runs substantially faster by
solving either the primal or dual (whichever has a smaller variable dimension)
of the QCLP.
| no_new_dataset | 0.947575 |
1502.07226 | Haifeng Wang | Haifeng Wang, R. Todd Constable and Gigi Galiana | Accelerate Single-shot Data Acquisitions Using Compressed Sensing and
FRONSAC Imaging | 4 pages, 4 figures, accepted to IEEE International Symposium on
Biomedical Imaging (ISBI) 2015 | null | 10.1109/ISBI.2015.7164101 | null | physics.med-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nonlinear spatial encoding magnetic (SEM) fields have been studied to
complement multichannel RF encoding and accelerate MRI scans. Published schemes
include PatLoc, O-Space, Null Space, 4D-RIO, and others, but the large variety
of possible approaches to exploiting nonlinear SEMs remains mostly unexplored.
Before, we have presented a new approach, Fast ROtary Nonlinear Spatial
ACquisition (FRONSAC) imaging, where the nonlinear fields provide a small
rotating perturbation to standard linear trajectories. While FRONSAC encoding
greatly improves image quality, at the highest accelerations or weakest FRONSAC
fields, some undersampling artifacts remain. However, the under-sampling
artifacts that occur with FRONSAC encoding are relatively incoherent and well
suited to the compressed sensing (CS) reconstruction. CS provides a
sparsity-promoting convex strategy to reconstruct images from highly
undersampled datasets. The work presented here combines the benefits of FRONSAC
and CS. Simulations illustrate that this combination can further improve image
reconstruction with FRONSAC gradients of low amplitudes and frequencies.
| [
{
"version": "v1",
"created": "Wed, 25 Feb 2015 16:21:23 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Apr 2015 20:42:19 GMT"
},
{
"version": "v3",
"created": "Fri, 24 Apr 2015 20:43:45 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Wang",
"Haifeng",
""
],
[
"Constable",
"R. Todd",
""
],
[
"Galiana",
"Gigi",
""
]
] | TITLE: Accelerate Single-shot Data Acquisitions Using Compressed Sensing and
FRONSAC Imaging
ABSTRACT: Nonlinear spatial encoding magnetic (SEM) fields have been studied to
complement multichannel RF encoding and accelerate MRI scans. Published schemes
include PatLoc, O-Space, Null Space, 4D-RIO, and others, but the large variety
of possible approaches to exploiting nonlinear SEMs remains mostly unexplored.
Before, we have presented a new approach, Fast ROtary Nonlinear Spatial
ACquisition (FRONSAC) imaging, where the nonlinear fields provide a small
rotating perturbation to standard linear trajectories. While FRONSAC encoding
greatly improves image quality, at the highest accelerations or weakest FRONSAC
fields, some undersampling artifacts remain. However, the under-sampling
artifacts that occur with FRONSAC encoding are relatively incoherent and well
suited to the compressed sensing (CS) reconstruction. CS provides a
sparsity-promoting convex strategy to reconstruct images from highly
undersampled datasets. The work presented here combines the benefits of FRONSAC
and CS. Simulations illustrate that this combination can further improve image
reconstruction with FRONSAC gradients of low amplitudes and frequencies.
| no_new_dataset | 0.948251 |
1503.00759 | Maximilian Nickel | Maximilian Nickel, Kevin Murphy, Volker Tresp, Evgeniy Gabrilovich | A Review of Relational Machine Learning for Knowledge Graphs | To appear in Proceedings of the IEEE | null | 10.1109/JPROC.2015.2483592 | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Relational machine learning studies methods for the statistical analysis of
relational, or graph-structured, data. In this paper, we provide a review of
how such statistical models can be "trained" on large knowledge graphs, and
then used to predict new facts about the world (which is equivalent to
predicting new edges in the graph). In particular, we discuss two fundamentally
different kinds of statistical relational models, both of which can scale to
massive datasets. The first is based on latent feature models such as tensor
factorization and multiway neural networks. The second is based on mining
observable patterns in the graph. We also show how to combine these latent and
observable models to get improved modeling power at decreased computational
cost. Finally, we discuss how such statistical models of graphs can be combined
with text-based information extraction methods for automatically constructing
knowledge graphs from the Web. To this end, we also discuss Google's Knowledge
Vault project as an example of such combination.
| [
{
"version": "v1",
"created": "Mon, 2 Mar 2015 21:35:41 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Aug 2015 16:35:31 GMT"
},
{
"version": "v3",
"created": "Mon, 28 Sep 2015 17:40:35 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Nickel",
"Maximilian",
""
],
[
"Murphy",
"Kevin",
""
],
[
"Tresp",
"Volker",
""
],
[
"Gabrilovich",
"Evgeniy",
""
]
] | TITLE: A Review of Relational Machine Learning for Knowledge Graphs
ABSTRACT: Relational machine learning studies methods for the statistical analysis of
relational, or graph-structured, data. In this paper, we provide a review of
how such statistical models can be "trained" on large knowledge graphs, and
then used to predict new facts about the world (which is equivalent to
predicting new edges in the graph). In particular, we discuss two fundamentally
different kinds of statistical relational models, both of which can scale to
massive datasets. The first is based on latent feature models such as tensor
factorization and multiway neural networks. The second is based on mining
observable patterns in the graph. We also show how to combine these latent and
observable models to get improved modeling power at decreased computational
cost. Finally, we discuss how such statistical models of graphs can be combined
with text-based information extraction methods for automatically constructing
knowledge graphs from the Web. To this end, we also discuss Google's Knowledge
Vault project as an example of such combination.
| no_new_dataset | 0.950503 |
1503.03832 | Florian Schroff | Florian Schroff, Dmitry Kalenichenko, James Philbin | FaceNet: A Unified Embedding for Face Recognition and Clustering | Also published, in Proceedings of the IEEE Computer Society
Conference on Computer Vision and Pattern Recognition 2015 | null | 10.1109/CVPR.2015.7298682 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite significant recent advances in the field of face recognition,
implementing face verification and recognition efficiently at scale presents
serious challenges to current approaches. In this paper we present a system,
called FaceNet, that directly learns a mapping from face images to a compact
Euclidean space where distances directly correspond to a measure of face
similarity. Once this space has been produced, tasks such as face recognition,
verification and clustering can be easily implemented using standard techniques
with FaceNet embeddings as feature vectors.
Our method uses a deep convolutional network trained to directly optimize the
embedding itself, rather than an intermediate bottleneck layer as in previous
deep learning approaches. To train, we use triplets of roughly aligned matching
/ non-matching face patches generated using a novel online triplet mining
method. The benefit of our approach is much greater representational
efficiency: we achieve state-of-the-art face recognition performance using only
128-bytes per face.
On the widely used Labeled Faces in the Wild (LFW) dataset, our system
achieves a new record accuracy of 99.63%. On YouTube Faces DB it achieves
95.12%. Our system cuts the error rate in comparison to the best published
result by 30% on both datasets.
We also introduce the concept of harmonic embeddings, and a harmonic triplet
loss, which describe different versions of face embeddings (produced by
different networks) that are compatible to each other and allow for direct
comparison between each other.
| [
{
"version": "v1",
"created": "Thu, 12 Mar 2015 18:10:53 GMT"
},
{
"version": "v2",
"created": "Thu, 4 Jun 2015 19:17:30 GMT"
},
{
"version": "v3",
"created": "Wed, 17 Jun 2015 23:35:47 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Schroff",
"Florian",
""
],
[
"Kalenichenko",
"Dmitry",
""
],
[
"Philbin",
"James",
""
]
] | TITLE: FaceNet: A Unified Embedding for Face Recognition and Clustering
ABSTRACT: Despite significant recent advances in the field of face recognition,
implementing face verification and recognition efficiently at scale presents
serious challenges to current approaches. In this paper we present a system,
called FaceNet, that directly learns a mapping from face images to a compact
Euclidean space where distances directly correspond to a measure of face
similarity. Once this space has been produced, tasks such as face recognition,
verification and clustering can be easily implemented using standard techniques
with FaceNet embeddings as feature vectors.
Our method uses a deep convolutional network trained to directly optimize the
embedding itself, rather than an intermediate bottleneck layer as in previous
deep learning approaches. To train, we use triplets of roughly aligned matching
/ non-matching face patches generated using a novel online triplet mining
method. The benefit of our approach is much greater representational
efficiency: we achieve state-of-the-art face recognition performance using only
128-bytes per face.
On the widely used Labeled Faces in the Wild (LFW) dataset, our system
achieves a new record accuracy of 99.63%. On YouTube Faces DB it achieves
95.12%. Our system cuts the error rate in comparison to the best published
result by 30% on both datasets.
We also introduce the concept of harmonic embeddings, and a harmonic triplet
loss, which describe different versions of face embeddings (produced by
different networks) that are compatible to each other and allow for direct
comparison between each other.
| no_new_dataset | 0.950503 |
1503.04877 | Syed Agha Muhammad | Syed Agha Muhammad and Kristof Van Laerhoven | An Automated System for Discovering Neighborhood Patterns in Ego
Networks | null | null | null | null | cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generally, social network analysis has often focused on the topology of the
network without considering the characteristics of individuals involved in
them. Less attention is given to study the behavior of individuals, considering
they are the basic entity of a graph. Given a mobile social network graph, what
are good features to extract key information from the nodes? How many distinct
neighborhood patterns exist for ego nodes? What clues does such information
provide to study nodes over a long period of time?
In this report, we develop an automated system in order to discover the
occurrences of prototypical ego-centric patterns from data. We aim to provide a
data-driven instrument to be used in behavioral sciences for graph
interpretations. We analyze social networks derived from real-world data
collected with smart-phones. We select 13 well-known network measures,
especially those concerned with ego graphs. We form eight feature subsets and
then assess their performance using unsupervised clustering techniques to
discover distinguishing ego-centric patterns. From clustering analysis, we
discover that eight distinct neighborhood patterns have emerged. This
categorization allows concise analysis of users' data as they change over time.
The results provide a fine-grained analysis for the contribution of different
feature sets to detect unique clustering patterns. Last, as a case study, two
datasets are studied over long periods to demonstrate the utility of this
method. The study shows the effectiveness of the proposed approach in
discovering important trends from data.
| [
{
"version": "v1",
"created": "Mon, 16 Mar 2015 23:07:12 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Muhammad",
"Syed Agha",
""
],
[
"Van Laerhoven",
"Kristof",
""
]
] | TITLE: An Automated System for Discovering Neighborhood Patterns in Ego
Networks
ABSTRACT: Generally, social network analysis has often focused on the topology of the
network without considering the characteristics of individuals involved in
them. Less attention is given to study the behavior of individuals, considering
they are the basic entity of a graph. Given a mobile social network graph, what
are good features to extract key information from the nodes? How many distinct
neighborhood patterns exist for ego nodes? What clues does such information
provide to study nodes over a long period of time?
In this report, we develop an automated system in order to discover the
occurrences of prototypical ego-centric patterns from data. We aim to provide a
data-driven instrument to be used in behavioral sciences for graph
interpretations. We analyze social networks derived from real-world data
collected with smart-phones. We select 13 well-known network measures,
especially those concerned with ego graphs. We form eight feature subsets and
then assess their performance using unsupervised clustering techniques to
discover distinguishing ego-centric patterns. From clustering analysis, we
discover that eight distinct neighborhood patterns have emerged. This
categorization allows concise analysis of users' data as they change over time.
The results provide a fine-grained analysis for the contribution of different
feature sets to detect unique clustering patterns. Last, as a case study, two
datasets are studied over long periods to demonstrate the utility of this
method. The study shows the effectiveness of the proposed approach in
discovering important trends from data.
| no_new_dataset | 0.9462 |
1503.08771 | Jinxue Zhang | Jinxue Zhang, Jingchao Sun, Rui Zhang, Yanchao Zhang | Your Actions Tell Where You Are: Uncovering Twitter Users in a
Metropolitan Area | Accepted by IEEE Conference on Communications and Network Security
(CNS) 2015 | null | 10.1109/CNS.2015.7346854 | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Twitter is an extremely popular social networking platform. Most Twitter
users do not disclose their locations due to privacy concerns. Although
inferring the location of an individual Twitter user has been extensively
studied, it is still missing to effectively find the majority of the users in a
specific geographical area without scanning the whole Twittersphere, and
obtaining these users will result in both positive and negative significance.
In this paper, we propose LocInfer, a novel and lightweight system to tackle
this problem. LocInfer explores the fact that user communications in Twitter
exhibit strong geographic locality, which we validate through large-scale
datasets. Based on the experiments from four representative metropolitan areas
in U.S., LocInfer can discover on average 86.6% of the users with 73.2%
accuracy in each area by only checking a small set of candidate users. We also
present a countermeasure to the users highly sensitive to location privacy and
show its efficacy by simulations.
| [
{
"version": "v1",
"created": "Mon, 30 Mar 2015 18:07:30 GMT"
},
{
"version": "v2",
"created": "Mon, 3 Aug 2015 23:46:19 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Zhang",
"Jinxue",
""
],
[
"Sun",
"Jingchao",
""
],
[
"Zhang",
"Rui",
""
],
[
"Zhang",
"Yanchao",
""
]
] | TITLE: Your Actions Tell Where You Are: Uncovering Twitter Users in a
Metropolitan Area
ABSTRACT: Twitter is an extremely popular social networking platform. Most Twitter
users do not disclose their locations due to privacy concerns. Although
inferring the location of an individual Twitter user has been extensively
studied, it is still missing to effectively find the majority of the users in a
specific geographical area without scanning the whole Twittersphere, and
obtaining these users will result in both positive and negative significance.
In this paper, we propose LocInfer, a novel and lightweight system to tackle
this problem. LocInfer explores the fact that user communications in Twitter
exhibit strong geographic locality, which we validate through large-scale
datasets. Based on the experiments from four representative metropolitan areas
in U.S., LocInfer can discover on average 86.6% of the users with 73.2%
accuracy in each area by only checking a small set of candidate users. We also
present a countermeasure to the users highly sensitive to location privacy and
show its efficacy by simulations.
| no_new_dataset | 0.946941 |
1504.03573 | Marcus A. Brubaker | Marcus A. Brubaker, Ali Punjani and David J. Fleet | Building Proteins in a Day: Efficient 3D Molecular Reconstruction | To be presented at IEEE Conference on Computer Vision and Pattern
Recognition (CVPR) 2015 | null | 10.1109/CVPR.2015.7298929 | null | cs.CV q-bio.QM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Discovering the 3D atomic structure of molecules such as proteins and viruses
is a fundamental research problem in biology and medicine. Electron
Cryomicroscopy (Cryo-EM) is a promising vision-based technique for structure
estimation which attempts to reconstruct 3D structures from 2D images. This
paper addresses the challenging problem of 3D reconstruction from 2D Cryo-EM
images. A new framework for estimation is introduced which relies on modern
stochastic optimization techniques to scale to large datasets. We also
introduce a novel technique which reduces the cost of evaluating the objective
function during optimization by over five orders or magnitude. The net result
is an approach capable of estimating 3D molecular structure from large scale
datasets in about a day on a single workstation.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 14:56:17 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Brubaker",
"Marcus A.",
""
],
[
"Punjani",
"Ali",
""
],
[
"Fleet",
"David J.",
""
]
] | TITLE: Building Proteins in a Day: Efficient 3D Molecular Reconstruction
ABSTRACT: Discovering the 3D atomic structure of molecules such as proteins and viruses
is a fundamental research problem in biology and medicine. Electron
Cryomicroscopy (Cryo-EM) is a promising vision-based technique for structure
estimation which attempts to reconstruct 3D structures from 2D images. This
paper addresses the challenging problem of 3D reconstruction from 2D Cryo-EM
images. A new framework for estimation is introduced which relies on modern
stochastic optimization techniques to scale to large datasets. We also
introduce a novel technique which reduces the cost of evaluating the objective
function during optimization by over five orders or magnitude. The net result
is an approach capable of estimating 3D molecular structure from large scale
datasets in about a day on a single workstation.
| no_new_dataset | 0.947769 |
1507.06838 | Modesto Castrill\'on-Santana | M. Castrill\'on-Santana, J. Lorenzo-Navarro and E. Ram\'on-Balmaseda | Descriptors and regions of interest fusion for gender classification in
the wild. Comparison and combination with Convolutional Neural Networks | Revised version containing 12 pages. This revision includes newer
referenes, results with CNN, fusion of local descriptors amd CNN and corrects
different typos | null | 10.1016/j.imavis.2016.10.004 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Gender classification (GC) has achieved high accuracy in different
experimental evaluations based mostly on inner facial details. However, these
results do not generalize well in unrestricted datasets and particularly in
cross-database experiments, where the performance drops drastically. In this
paper, we analyze the state-of-the-art GC accuracy on three large datasets:
MORPH, LFW and GROUPS. We discuss their respective difficulties and bias,
concluding that the most challenging and wildest complexity is present in
GROUPS. This dataset covers hard conditions such as low resolution imagery and
cluttered background. Firstly, we analyze in depth the performance of different
descriptors extracted from the face and its local context on this dataset.
Selecting the bests and studying their most suitable combination allows us to
design a solution that beats any previously published results for GROUPS with
the Dago's protocol, reaching an accuracy over 94.2%, reducing the gap with
other simpler datasets. The chosen solution based on local descriptors is later
evaluated in a cross-database scenario with the three mentioned datasets, and
full dataset 5-fold cross validation. The achieved results are compared with a
Convolutional Neural Network approach, achieving rather similar marks. Finally,
a solution is proposed combining both focuses, exhibiting great
complementarity, boosting GC performance to beat previously published results
in GC both cross-database, and full in-database evaluations.
| [
{
"version": "v1",
"created": "Fri, 24 Jul 2015 13:22:29 GMT"
},
{
"version": "v2",
"created": "Fri, 19 Feb 2016 12:44:13 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Castrillón-Santana",
"M.",
""
],
[
"Lorenzo-Navarro",
"J.",
""
],
[
"Ramón-Balmaseda",
"E.",
""
]
] | TITLE: Descriptors and regions of interest fusion for gender classification in
the wild. Comparison and combination with Convolutional Neural Networks
ABSTRACT: Gender classification (GC) has achieved high accuracy in different
experimental evaluations based mostly on inner facial details. However, these
results do not generalize well in unrestricted datasets and particularly in
cross-database experiments, where the performance drops drastically. In this
paper, we analyze the state-of-the-art GC accuracy on three large datasets:
MORPH, LFW and GROUPS. We discuss their respective difficulties and bias,
concluding that the most challenging and wildest complexity is present in
GROUPS. This dataset covers hard conditions such as low resolution imagery and
cluttered background. Firstly, we analyze in depth the performance of different
descriptors extracted from the face and its local context on this dataset.
Selecting the bests and studying their most suitable combination allows us to
design a solution that beats any previously published results for GROUPS with
the Dago's protocol, reaching an accuracy over 94.2%, reducing the gap with
other simpler datasets. The chosen solution based on local descriptors is later
evaluated in a cross-database scenario with the three mentioned datasets, and
full dataset 5-fold cross validation. The achieved results are compared with a
Convolutional Neural Network approach, achieving rather similar marks. Finally,
a solution is proposed combining both focuses, exhibiting great
complementarity, boosting GC performance to beat previously published results
in GC both cross-database, and full in-database evaluations.
| no_new_dataset | 0.949529 |
1511.04110 | Ali Mollahosseini | Ali Mollahosseini, David Chan, Mohammad H. Mahoor | Going Deeper in Facial Expression Recognition using Deep Neural Networks | To be appear in IEEE Winter Conference on Applications of Computer
Vision (WACV), 2016 {Accepted in first round submission} | IEEE Winter Conference on Applications of Computer Vision (WACV),
2016 | 10.1109/WACV.2016.7477450 | null | cs.NE cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automated Facial Expression Recognition (FER) has remained a challenging and
interesting problem. Despite efforts made in developing various methods for
FER, existing approaches traditionally lack generalizability when applied to
unseen images or those that are captured in wild setting. Most of the existing
approaches are based on engineered features (e.g. HOG, LBPH, and Gabor) where
the classifier's hyperparameters are tuned to give best recognition accuracies
across a single database, or a small collection of similar databases.
Nevertheless, the results are not significant when they are applied to novel
data. This paper proposes a deep neural network architecture to address the FER
problem across multiple well-known standard face datasets. Specifically, our
network consists of two convolutional layers each followed by max pooling and
then four Inception layers. The network is a single component architecture that
takes registered facial images as the input and classifies them into either of
the six basic or the neutral expressions. We conducted comprehensive
experiments on seven publically available facial expression databases, viz.
MultiPIE, MMI, CK+, DISFA, FERA, SFEW, and FER2013. The results of proposed
architecture are comparable to or better than the state-of-the-art methods and
better than traditional convolutional neural networks and in both accuracy and
training time.
| [
{
"version": "v1",
"created": "Thu, 12 Nov 2015 22:10:46 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Mollahosseini",
"Ali",
""
],
[
"Chan",
"David",
""
],
[
"Mahoor",
"Mohammad H.",
""
]
] | TITLE: Going Deeper in Facial Expression Recognition using Deep Neural Networks
ABSTRACT: Automated Facial Expression Recognition (FER) has remained a challenging and
interesting problem. Despite efforts made in developing various methods for
FER, existing approaches traditionally lack generalizability when applied to
unseen images or those that are captured in wild setting. Most of the existing
approaches are based on engineered features (e.g. HOG, LBPH, and Gabor) where
the classifier's hyperparameters are tuned to give best recognition accuracies
across a single database, or a small collection of similar databases.
Nevertheless, the results are not significant when they are applied to novel
data. This paper proposes a deep neural network architecture to address the FER
problem across multiple well-known standard face datasets. Specifically, our
network consists of two convolutional layers each followed by max pooling and
then four Inception layers. The network is a single component architecture that
takes registered facial images as the input and classifies them into either of
the six basic or the neutral expressions. We conducted comprehensive
experiments on seven publically available facial expression databases, viz.
MultiPIE, MMI, CK+, DISFA, FERA, SFEW, and FER2013. The results of proposed
architecture are comparable to or better than the state-of-the-art methods and
better than traditional convolutional neural networks and in both accuracy and
training time.
| no_new_dataset | 0.949902 |
1511.04192 | Tianshui Chen | Tianshui Chen, Liang Lin, Lingbo Liu, Xiaonan Luo, Xuelong Li | DISC: Deep Image Saliency Computing via Progressive Representation
Learning | This manuscript is the accepted version for IEEE Transactions on
Neural Networks and Learning Systems (T-NNLS), 2015 | null | 10.1109/TNNLS.2015.2506664 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Salient object detection increasingly receives attention as an important
component or step in several pattern recognition and image processing tasks.
Although a variety of powerful saliency models have been intensively proposed,
they usually involve heavy feature (or model) engineering based on priors (or
assumptions) about the properties of objects and backgrounds. Inspired by the
effectiveness of recently developed feature learning, we provide a novel Deep
Image Saliency Computing (DISC) framework for fine-grained image saliency
computing. In particular, we model the image saliency from both the coarse- and
fine-level observations, and utilize the deep convolutional neural network
(CNN) to learn the saliency representation in a progressive manner.
Specifically, our saliency model is built upon two stacked CNNs. The first CNN
generates a coarse-level saliency map by taking the overall image as the input,
roughly identifying saliency regions in the global context. Furthermore, we
integrate superpixel-based local context information in the first CNN to refine
the coarse-level saliency map. Guided by the coarse saliency map, the second
CNN focuses on the local context to produce fine-grained and accurate saliency
map while preserving object details. For a testing image, the two CNNs
collaboratively conduct the saliency computing in one shot. Our DISC framework
is capable of uniformly highlighting the objects-of-interest from complex
background while preserving well object details. Extensive experiments on
several standard benchmarks suggest that DISC outperforms other
state-of-the-art methods and it also generalizes well across datasets without
additional training. The executable version of DISC is available online:
http://vision.sysu.edu.cn/projects/DISC.
| [
{
"version": "v1",
"created": "Fri, 13 Nov 2015 07:14:13 GMT"
},
{
"version": "v2",
"created": "Thu, 10 Dec 2015 13:11:23 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Chen",
"Tianshui",
""
],
[
"Lin",
"Liang",
""
],
[
"Liu",
"Lingbo",
""
],
[
"Luo",
"Xiaonan",
""
],
[
"Li",
"Xuelong",
""
]
] | TITLE: DISC: Deep Image Saliency Computing via Progressive Representation
Learning
ABSTRACT: Salient object detection increasingly receives attention as an important
component or step in several pattern recognition and image processing tasks.
Although a variety of powerful saliency models have been intensively proposed,
they usually involve heavy feature (or model) engineering based on priors (or
assumptions) about the properties of objects and backgrounds. Inspired by the
effectiveness of recently developed feature learning, we provide a novel Deep
Image Saliency Computing (DISC) framework for fine-grained image saliency
computing. In particular, we model the image saliency from both the coarse- and
fine-level observations, and utilize the deep convolutional neural network
(CNN) to learn the saliency representation in a progressive manner.
Specifically, our saliency model is built upon two stacked CNNs. The first CNN
generates a coarse-level saliency map by taking the overall image as the input,
roughly identifying saliency regions in the global context. Furthermore, we
integrate superpixel-based local context information in the first CNN to refine
the coarse-level saliency map. Guided by the coarse saliency map, the second
CNN focuses on the local context to produce fine-grained and accurate saliency
map while preserving object details. For a testing image, the two CNNs
collaboratively conduct the saliency computing in one shot. Our DISC framework
is capable of uniformly highlighting the objects-of-interest from complex
background while preserving well object details. Extensive experiments on
several standard benchmarks suggest that DISC outperforms other
state-of-the-art methods and it also generalizes well across datasets without
additional training. The executable version of DISC is available online:
http://vision.sysu.edu.cn/projects/DISC.
| no_new_dataset | 0.951051 |
1602.06888 | Maria Patterson | Maria T Patterson, Nikolas Anderson, Collin Bennett, Jacob Bruggemann,
Robert Grossman, Matthew Handy, Vuong Ly, Dan Mandl, Shane Pederson, Jim
Pivarski, Ray Powell, Jonathan Spring and Walt Wells | The Matsu Wheel: A Cloud-based Framework for Efficient Analysis and
Reanalysis of Earth Satellite Imagery | 10 pages, accepted for presentation to IEEE BigDataService 2016 | null | 10.1109/BigDataService.2016.39 | null | cs.DC astro-ph.IM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Project Matsu is a collaboration between the Open Commons Consortium and NASA
focused on developing open source technology for the cloud-based processing of
Earth satellite imagery. A particular focus is the development of applications
for detecting fires and floods to help support natural disaster detection and
relief. Project Matsu has developed an open source cloud-based infrastructure
to process, analyze, and reanalyze large collections of hyperspectral satellite
image data using OpenStack, Hadoop, MapReduce, Storm and related technologies.
We describe a framework for efficient analysis of large amounts of data
called the Matsu "Wheel." The Matsu Wheel is currently used to process incoming
hyperspectral satellite data produced daily by NASA's Earth Observing-1 (EO-1)
satellite. The framework is designed to be able to support scanning queries
using cloud computing applications, such as Hadoop and Accumulo. A scanning
query processes all, or most of the data, in a database or data repository.
We also describe our preliminary Wheel analytics, including an anomaly
detector for rare spectral signatures or thermal anomalies in hyperspectral
data and a land cover classifier that can be used for water and flood
detection. Each of these analytics can generate visual reports accessible via
the web for the public and interested decision makers. The resultant products
of the analytics are also made accessible through an Open Geospatial Compliant
(OGC)-compliant Web Map Service (WMS) for further distribution. The Matsu Wheel
allows many shared data services to be performed together to efficiently use
resources for processing hyperspectral satellite image data and other, e.g.,
large environmental datasets that may be analyzed for many purposes.
| [
{
"version": "v1",
"created": "Mon, 22 Feb 2016 18:51:14 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Patterson",
"Maria T",
""
],
[
"Anderson",
"Nikolas",
""
],
[
"Bennett",
"Collin",
""
],
[
"Bruggemann",
"Jacob",
""
],
[
"Grossman",
"Robert",
""
],
[
"Handy",
"Matthew",
""
],
[
"Ly",
"Vuong",
""
],
[
"Mandl",
"Dan",
""
],
[
"Pederson",
"Shane",
""
],
[
"Pivarski",
"Jim",
""
],
[
"Powell",
"Ray",
""
],
[
"Spring",
"Jonathan",
""
],
[
"Wells",
"Walt",
""
]
] | TITLE: The Matsu Wheel: A Cloud-based Framework for Efficient Analysis and
Reanalysis of Earth Satellite Imagery
ABSTRACT: Project Matsu is a collaboration between the Open Commons Consortium and NASA
focused on developing open source technology for the cloud-based processing of
Earth satellite imagery. A particular focus is the development of applications
for detecting fires and floods to help support natural disaster detection and
relief. Project Matsu has developed an open source cloud-based infrastructure
to process, analyze, and reanalyze large collections of hyperspectral satellite
image data using OpenStack, Hadoop, MapReduce, Storm and related technologies.
We describe a framework for efficient analysis of large amounts of data
called the Matsu "Wheel." The Matsu Wheel is currently used to process incoming
hyperspectral satellite data produced daily by NASA's Earth Observing-1 (EO-1)
satellite. The framework is designed to be able to support scanning queries
using cloud computing applications, such as Hadoop and Accumulo. A scanning
query processes all, or most of the data, in a database or data repository.
We also describe our preliminary Wheel analytics, including an anomaly
detector for rare spectral signatures or thermal anomalies in hyperspectral
data and a land cover classifier that can be used for water and flood
detection. Each of these analytics can generate visual reports accessible via
the web for the public and interested decision makers. The resultant products
of the analytics are also made accessible through an Open Geospatial Compliant
(OGC)-compliant Web Map Service (WMS) for further distribution. The Matsu Wheel
allows many shared data services to be performed together to efficiently use
resources for processing hyperspectral satellite image data and other, e.g.,
large environmental datasets that may be analyzed for many purposes.
| no_new_dataset | 0.959269 |
1602.09076 | Paolo Campigotto | Paolo Campigotto, Christian Rudloff, Maximilian Leodolter and Dietmar
Bauer | Personalized and situation-aware multimodal route recommendations: the
FAVOUR algorithm | 12 pages, 6 figures, 1 table. Submitted to IEEE Transactions on
Intelligent Transportation Systems journal for publication | null | 10.1109/TITS.2016.2565643 | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Route choice in multimodal networks shows a considerable variation between
different individuals as well as the current situational context.
Personalization of recommendation algorithms are already common in many areas,
e.g., online retail. However, most online routing applications still provide
shortest distance or shortest travel-time routes only, neglecting individual
preferences as well as the current situation. Both aspects are of particular
importance in a multimodal setting as attractivity of some transportation modes
such as biking crucially depends on personal characteristics and exogenous
factors like the weather. This paper introduces the FAVourite rOUte
Recommendation (FAVOUR) approach to provide personalized, situation-aware route
proposals based on three steps: first, at the initialization stage, the user
provides limited information (home location, work place, mobility options,
sociodemographics) used to select one out of a small number of initial
profiles. Second, based on this information, a stated preference survey is
designed in order to sharpen the profile. In this step a mass preference prior
is used to encode the prior knowledge on preferences from the class identified
in step one. And third, subsequently the profile is continuously updated during
usage of the routing services. The last two steps use Bayesian learning
techniques in order to incorporate information from all contributing
individuals. The FAVOUR approach is presented in detail and tested on a small
number of survey participants. The experimental results on this real-world
dataset show that FAVOUR generates better-quality recommendations w.r.t.
alternative learning algorithms from the literature. In particular the
definition of the mass preference prior for initialization of step two is shown
to provide better predictions than a number of alternatives from the
literature.
| [
{
"version": "v1",
"created": "Mon, 29 Feb 2016 18:16:12 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Campigotto",
"Paolo",
""
],
[
"Rudloff",
"Christian",
""
],
[
"Leodolter",
"Maximilian",
""
],
[
"Bauer",
"Dietmar",
""
]
] | TITLE: Personalized and situation-aware multimodal route recommendations: the
FAVOUR algorithm
ABSTRACT: Route choice in multimodal networks shows a considerable variation between
different individuals as well as the current situational context.
Personalization of recommendation algorithms are already common in many areas,
e.g., online retail. However, most online routing applications still provide
shortest distance or shortest travel-time routes only, neglecting individual
preferences as well as the current situation. Both aspects are of particular
importance in a multimodal setting as attractivity of some transportation modes
such as biking crucially depends on personal characteristics and exogenous
factors like the weather. This paper introduces the FAVourite rOUte
Recommendation (FAVOUR) approach to provide personalized, situation-aware route
proposals based on three steps: first, at the initialization stage, the user
provides limited information (home location, work place, mobility options,
sociodemographics) used to select one out of a small number of initial
profiles. Second, based on this information, a stated preference survey is
designed in order to sharpen the profile. In this step a mass preference prior
is used to encode the prior knowledge on preferences from the class identified
in step one. And third, subsequently the profile is continuously updated during
usage of the routing services. The last two steps use Bayesian learning
techniques in order to incorporate information from all contributing
individuals. The FAVOUR approach is presented in detail and tested on a small
number of survey participants. The experimental results on this real-world
dataset show that FAVOUR generates better-quality recommendations w.r.t.
alternative learning algorithms from the literature. In particular the
definition of the mass preference prior for initialization of step two is shown
to provide better predictions than a number of alternatives from the
literature.
| no_new_dataset | 0.953622 |
1604.07807 | Shangxuan Wu | Shangxuan Wu, Ying-Cong Chen, Xiang Li, An-Cong Wu, Jin-Jie You, and
Wei-Shi Zheng | An Enhanced Deep Feature Representation for Person Re-identification | Citation for this paper: Shangxuan Wu, Ying-Cong Chen, Xiang Li,
An-Cong Wu, Jin-Jie You, and Wei-Shi Zheng. An Enhanced Deep Feature
Representation for Person Re-identification. In IEEE WACV, 2016 | null | 10.1109/WACV.2016.7477681 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Feature representation and metric learning are two critical components in
person re-identification models. In this paper, we focus on the feature
representation and claim that hand-crafted histogram features can be
complementary to Convolutional Neural Network (CNN) features. We propose a
novel feature extraction model called Feature Fusion Net (FFN) for pedestrian
image representation. In FFN, back propagation makes CNN features constrained
by the handcrafted features. Utilizing color histogram features (RGB, HSV,
YCbCr, Lab and YIQ) and texture features (multi-scale and multi-orientation
Gabor features), we get a new deep feature representation that is more
discriminative and compact. Experiments on three challenging datasets (VIPeR,
CUHK01, PRID450s) validates the effectiveness of our proposal.
| [
{
"version": "v1",
"created": "Tue, 26 Apr 2016 19:27:50 GMT"
},
{
"version": "v2",
"created": "Thu, 28 Apr 2016 08:17:35 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Wu",
"Shangxuan",
""
],
[
"Chen",
"Ying-Cong",
""
],
[
"Li",
"Xiang",
""
],
[
"Wu",
"An-Cong",
""
],
[
"You",
"Jin-Jie",
""
],
[
"Zheng",
"Wei-Shi",
""
]
] | TITLE: An Enhanced Deep Feature Representation for Person Re-identification
ABSTRACT: Feature representation and metric learning are two critical components in
person re-identification models. In this paper, we focus on the feature
representation and claim that hand-crafted histogram features can be
complementary to Convolutional Neural Network (CNN) features. We propose a
novel feature extraction model called Feature Fusion Net (FFN) for pedestrian
image representation. In FFN, back propagation makes CNN features constrained
by the handcrafted features. Utilizing color histogram features (RGB, HSV,
YCbCr, Lab and YIQ) and texture features (multi-scale and multi-orientation
Gabor features), we get a new deep feature representation that is more
discriminative and compact. Experiments on three challenging datasets (VIPeR,
CUHK01, PRID450s) validates the effectiveness of our proposal.
| no_new_dataset | 0.947186 |
1606.01593 | Steve Versteeg | Jean-Guy Schneider, Peter Mandile, Steve Versteeg | Generalized Suffix Tree based Multiple Sequence Alignment for Service
Virtualization | In Proceedings of 24th Australasian Software Engineering Conference
(ASWEC2015), pp 48-57 | In Proceedings of 24th Australasian Software Engineering
Conference (ASWEC2015), pp 48-57. IEEE | 10.1109/ASWEC.2015.13 | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Assuring quality of contemporary software systems is a very challenging task
due to the often large complexity of the deployment environments in which they
will operate. Service virtualization is an approach to this challenge where
services within the deployment environment are emulated by synthesising service
response messages from models or by recording and then replaying service
interaction messages with the system. Record-and-replay techniques require an
approach where (i) message prototypes can be derived from recorded system
interactions (i.e. request-response sequences), (ii) a scheme to match incoming
request messages against message prototypes, and (iii) the synthesis of
response messages based on similarities between incoming messages and the
recorded system interactions. Previous approaches in service virtualization
have required a multiple sequence alignment (MSA) algorithm as a means of
finding common patterns of similarities and differences between messages
required by all three steps.
In this paper, we present a novel MSA algorithm based on Generalized Suffix
Trees (GSTs). We evaluated the accuracy and efficiency of the proposed
algorithm against six enterprise service message trace datasets, with the
proposed algorithm performing up to 50 times faster than standard MSA
approaches. Furthermore, the algorithm has applicability to other domains
beyond service virtualization.
| [
{
"version": "v1",
"created": "Mon, 6 Jun 2016 01:45:59 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Schneider",
"Jean-Guy",
""
],
[
"Mandile",
"Peter",
""
],
[
"Versteeg",
"Steve",
""
]
] | TITLE: Generalized Suffix Tree based Multiple Sequence Alignment for Service
Virtualization
ABSTRACT: Assuring quality of contemporary software systems is a very challenging task
due to the often large complexity of the deployment environments in which they
will operate. Service virtualization is an approach to this challenge where
services within the deployment environment are emulated by synthesising service
response messages from models or by recording and then replaying service
interaction messages with the system. Record-and-replay techniques require an
approach where (i) message prototypes can be derived from recorded system
interactions (i.e. request-response sequences), (ii) a scheme to match incoming
request messages against message prototypes, and (iii) the synthesis of
response messages based on similarities between incoming messages and the
recorded system interactions. Previous approaches in service virtualization
have required a multiple sequence alignment (MSA) algorithm as a means of
finding common patterns of similarities and differences between messages
required by all three steps.
In this paper, we present a novel MSA algorithm based on Generalized Suffix
Trees (GSTs). We evaluated the accuracy and efficiency of the proposed
algorithm against six enterprise service message trace datasets, with the
proposed algorithm performing up to 50 times faster than standard MSA
approaches. Furthermore, the algorithm has applicability to other domains
beyond service virtualization.
| no_new_dataset | 0.941654 |
1606.06259 | Amir Zadeh | Amir Zadeh, Rowan Zellers, Eli Pincus, Louis-Philippe Morency | MOSI: Multimodal Corpus of Sentiment Intensity and Subjectivity Analysis
in Online Opinion Videos | Accepted as Journal Publication in IEEE Intelligent Systems | IEEE Intelligent Systems 31.6 (2016): 82-88 | null | null | cs.CL cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | People are sharing their opinions, stories and reviews through online video
sharing websites every day. Studying sentiment and subjectivity in these
opinion videos is experiencing a growing attention from academia and industry.
While sentiment analysis has been successful for text, it is an understudied
research question for videos and multimedia content. The biggest setbacks for
studies in this direction are lack of a proper dataset, methodology, baselines
and statistical analysis of how information from different modality sources
relate to each other. This paper introduces to the scientific community the
first opinion-level annotated corpus of sentiment and subjectivity analysis in
online videos called Multimodal Opinion-level Sentiment Intensity dataset
(MOSI). The dataset is rigorously annotated with labels for subjectivity,
sentiment intensity, per-frame and per-opinion annotated visual features, and
per-milliseconds annotated audio features. Furthermore, we present baselines
for future studies in this direction as well as a new multimodal fusion
approach that jointly models spoken words and visual gestures.
| [
{
"version": "v1",
"created": "Mon, 20 Jun 2016 19:23:53 GMT"
},
{
"version": "v2",
"created": "Fri, 12 Aug 2016 02:39:40 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Zadeh",
"Amir",
""
],
[
"Zellers",
"Rowan",
""
],
[
"Pincus",
"Eli",
""
],
[
"Morency",
"Louis-Philippe",
""
]
] | TITLE: MOSI: Multimodal Corpus of Sentiment Intensity and Subjectivity Analysis
in Online Opinion Videos
ABSTRACT: People are sharing their opinions, stories and reviews through online video
sharing websites every day. Studying sentiment and subjectivity in these
opinion videos is experiencing a growing attention from academia and industry.
While sentiment analysis has been successful for text, it is an understudied
research question for videos and multimedia content. The biggest setbacks for
studies in this direction are lack of a proper dataset, methodology, baselines
and statistical analysis of how information from different modality sources
relate to each other. This paper introduces to the scientific community the
first opinion-level annotated corpus of sentiment and subjectivity analysis in
online videos called Multimodal Opinion-level Sentiment Intensity dataset
(MOSI). The dataset is rigorously annotated with labels for subjectivity,
sentiment intensity, per-frame and per-opinion annotated visual features, and
per-milliseconds annotated audio features. Furthermore, we present baselines
for future studies in this direction as well as a new multimodal fusion
approach that jointly models spoken words and visual gestures.
| new_dataset | 0.956513 |
1606.08150 | Da Li | Hancheng Wu, Da Li, Michela Becchi | Compiler-Assisted Workload Consolidation For Efficient Dynamic
Parallelism on GPU | 10 pages, 10 figures, to be published in IEEE International Parallel
and Distributed Processing Symposium (IPDPS) 2016 | null | 10.1109/IPDPS.2016.98 | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | GPUs have been widely used to accelerate computations exhibiting simple
patterns of parallelism - such as flat or two-level parallelism - and a degree
of parallelism that can be statically determined based on the size of the input
dataset. However, the effective use of GPUs for algorithms exhibiting complex
patterns of parallelism, possibly known only at runtime, is still an open
problem. Recently, Nvidia has introduced Dynamic Parallelism (DP) in its GPUs.
By making it possible to launch kernels directly from GPU threads, this feature
enables nested parallelism at runtime. However, the effective use of DP must
still be understood: a naive use of this feature may suffer from significant
runtime overhead and lead to GPU underutilization, resulting in poor
performance. In this work, we target this problem. First, we demonstrate how a
naive use of DP can result in poor performance. Second, we propose three
workload consolidation schemes to improve performance and hardware utilization
of DP-based codes, and we implement these code transformations in a
directive-based compiler. Finally, we evaluate our framework on two categories
of applications: algorithms including irregular loops and algorithms exhibiting
parallel recursion. Our experiments show that our approach significantly
reduces runtime overhead and improves GPU utilization, leading to speedup
factors from 90x to 3300x over basic DP-based solutions and speedups from 2x to
6x over flat implementations.
| [
{
"version": "v1",
"created": "Mon, 27 Jun 2016 07:55:23 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Wu",
"Hancheng",
""
],
[
"Li",
"Da",
""
],
[
"Becchi",
"Michela",
""
]
] | TITLE: Compiler-Assisted Workload Consolidation For Efficient Dynamic
Parallelism on GPU
ABSTRACT: GPUs have been widely used to accelerate computations exhibiting simple
patterns of parallelism - such as flat or two-level parallelism - and a degree
of parallelism that can be statically determined based on the size of the input
dataset. However, the effective use of GPUs for algorithms exhibiting complex
patterns of parallelism, possibly known only at runtime, is still an open
problem. Recently, Nvidia has introduced Dynamic Parallelism (DP) in its GPUs.
By making it possible to launch kernels directly from GPU threads, this feature
enables nested parallelism at runtime. However, the effective use of DP must
still be understood: a naive use of this feature may suffer from significant
runtime overhead and lead to GPU underutilization, resulting in poor
performance. In this work, we target this problem. First, we demonstrate how a
naive use of DP can result in poor performance. Second, we propose three
workload consolidation schemes to improve performance and hardware utilization
of DP-based codes, and we implement these code transformations in a
directive-based compiler. Finally, we evaluate our framework on two categories
of applications: algorithms including irregular loops and algorithms exhibiting
parallel recursion. Our experiments show that our approach significantly
reduces runtime overhead and improves GPU utilization, leading to speedup
factors from 90x to 3300x over basic DP-based solutions and speedups from 2x to
6x over flat implementations.
| no_new_dataset | 0.941708 |
1608.07605 | Cem Aksoylar | Cem Aksoylar, Jing Qian, Venkatesh Saligrama | Clustering and Community Detection with Imbalanced Clusters | Extended version of arXiv:1309.2303 with new applications. Accepted
to IEEE TSIPN | null | 10.1109/TSIPN.2016.2601022 | null | stat.ML cs.LG cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Spectral clustering methods which are frequently used in clustering and
community detection applications are sensitive to the specific graph
constructions particularly when imbalanced clusters are present. We show that
ratio cut (RCut) or normalized cut (NCut) objectives are not tailored to
imbalanced cluster sizes since they tend to emphasize cut sizes over cut
values. We propose a graph partitioning problem that seeks minimum cut
partitions under minimum size constraints on partitions to deal with imbalanced
cluster sizes. Our approach parameterizes a family of graphs by adaptively
modulating node degrees on a fixed node set, yielding a set of parameter
dependent cuts reflecting varying levels of imbalance. The solution to our
problem is then obtained by optimizing over these parameters. We present
rigorous limit cut analysis results to justify our approach and demonstrate the
superiority of our method through experiments on synthetic and real datasets
for data clustering, semi-supervised learning and community detection.
| [
{
"version": "v1",
"created": "Fri, 26 Aug 2016 20:49:06 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Aksoylar",
"Cem",
""
],
[
"Qian",
"Jing",
""
],
[
"Saligrama",
"Venkatesh",
""
]
] | TITLE: Clustering and Community Detection with Imbalanced Clusters
ABSTRACT: Spectral clustering methods which are frequently used in clustering and
community detection applications are sensitive to the specific graph
constructions particularly when imbalanced clusters are present. We show that
ratio cut (RCut) or normalized cut (NCut) objectives are not tailored to
imbalanced cluster sizes since they tend to emphasize cut sizes over cut
values. We propose a graph partitioning problem that seeks minimum cut
partitions under minimum size constraints on partitions to deal with imbalanced
cluster sizes. Our approach parameterizes a family of graphs by adaptively
modulating node degrees on a fixed node set, yielding a set of parameter
dependent cuts reflecting varying levels of imbalance. The solution to our
problem is then obtained by optimizing over these parameters. We present
rigorous limit cut analysis results to justify our approach and demonstrate the
superiority of our method through experiments on synthetic and real datasets
for data clustering, semi-supervised learning and community detection.
| no_new_dataset | 0.949482 |
1610.04211 | Julien Perez | Julien Perez and Fei Liu | Gated End-to-End Memory Networks | 9 pages, 3 figures, 3 tables | null | null | null | cs.CL stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Machine reading using differentiable reasoning models has recently shown
remarkable progress. In this context, End-to-End trainable Memory Networks,
MemN2N, have demonstrated promising performance on simple natural language
based reasoning tasks such as factual reasoning and basic deduction. However,
other tasks, namely multi-fact question-answering, positional reasoning or
dialog related tasks, remain challenging particularly due to the necessity of
more complex interactions between the memory and controller modules composing
this family of models. In this paper, we introduce a novel end-to-end memory
access regulation mechanism inspired by the current progress on the connection
short-cutting principle in the field of computer vision. Concretely, we develop
a Gated End-to-End trainable Memory Network architecture, GMemN2N. From the
machine learning perspective, this new capability is learned in an end-to-end
fashion without the use of any additional supervision signal which is, as far
as our knowledge goes, the first of its kind. Our experiments show significant
improvements on the most challenging tasks in the 20 bAbI dataset, without the
use of any domain knowledge. Then, we show improvements on the dialog bAbI
tasks including the real human-bot conversion-based Dialog State Tracking
Challenge (DSTC-2) dataset. On these two datasets, our model sets the new state
of the art.
| [
{
"version": "v1",
"created": "Thu, 13 Oct 2016 19:38:03 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Nov 2016 15:09:29 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Perez",
"Julien",
""
],
[
"Liu",
"Fei",
""
]
] | TITLE: Gated End-to-End Memory Networks
ABSTRACT: Machine reading using differentiable reasoning models has recently shown
remarkable progress. In this context, End-to-End trainable Memory Networks,
MemN2N, have demonstrated promising performance on simple natural language
based reasoning tasks such as factual reasoning and basic deduction. However,
other tasks, namely multi-fact question-answering, positional reasoning or
dialog related tasks, remain challenging particularly due to the necessity of
more complex interactions between the memory and controller modules composing
this family of models. In this paper, we introduce a novel end-to-end memory
access regulation mechanism inspired by the current progress on the connection
short-cutting principle in the field of computer vision. Concretely, we develop
a Gated End-to-End trainable Memory Network architecture, GMemN2N. From the
machine learning perspective, this new capability is learned in an end-to-end
fashion without the use of any additional supervision signal which is, as far
as our knowledge goes, the first of its kind. Our experiments show significant
improvements on the most challenging tasks in the 20 bAbI dataset, without the
use of any domain knowledge. Then, we show improvements on the dialog bAbI
tasks including the real human-bot conversion-based Dialog State Tracking
Challenge (DSTC-2) dataset. On these two datasets, our model sets the new state
of the art.
| no_new_dataset | 0.93744 |
1611.05340 | Abhay Gupta | Abhay Gupta | Approximating Wisdom of Crowds using K-RBMs | 8 pages, 1 figure | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | An important way to make large training sets is to gather noisy labels from
crowds of non experts. We propose a method to aggregate noisy labels collected
from a crowd of workers or annotators. Eliciting labels is important in tasks
such as judging web search quality and rating products. Our method assumes that
labels are generated by a probability distribution over items and labels. We
formulate the method by drawing parallels between Gaussian Mixture Models
(GMMs) and Restricted Boltzmann Machines (RBMs) and show that the problem of
vote aggregation can be viewed as one of clustering. We use K-RBMs to perform
clustering. We finally show some empirical evaluations over real datasets.
| [
{
"version": "v1",
"created": "Wed, 16 Nov 2016 16:01:48 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Nov 2016 02:48:04 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Gupta",
"Abhay",
""
]
] | TITLE: Approximating Wisdom of Crowds using K-RBMs
ABSTRACT: An important way to make large training sets is to gather noisy labels from
crowds of non experts. We propose a method to aggregate noisy labels collected
from a crowd of workers or annotators. Eliciting labels is important in tasks
such as judging web search quality and rating products. Our method assumes that
labels are generated by a probability distribution over items and labels. We
formulate the method by drawing parallels between Gaussian Mixture Models
(GMMs) and Restricted Boltzmann Machines (RBMs) and show that the problem of
vote aggregation can be viewed as one of clustering. We use K-RBMs to perform
clustering. We finally show some empirical evaluations over real datasets.
| no_new_dataset | 0.94699 |
1611.05503 | Yu Liu | Yu Liu, Yanming Guo, and Michael S. Lew | On the Exploration of Convolutional Fusion Networks for Visual
Recognition | 23rd International Conference on MultiMedia Modeling (MMM 2017) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite recent advances in multi-scale deep representations, their
limitations are attributed to expensive parameters and weak fusion modules.
Hence, we propose an efficient approach to fuse multi-scale deep
representations, called convolutional fusion networks (CFN). Owing to using
1$\times$1 convolution and global average pooling, CFN can efficiently generate
the side branches while adding few parameters. In addition, we present a
locally-connected fusion module, which can learn adaptive weights for the side
branches and form a discriminatively fused feature. CFN models trained on the
CIFAR and ImageNet datasets demonstrate remarkable improvements over the plain
CNNs. Furthermore, we generalize CFN to three new tasks, including scene
recognition, fine-grained recognition and image retrieval. Our experiments show
that it can obtain consistent improvements towards the transferring tasks.
| [
{
"version": "v1",
"created": "Wed, 16 Nov 2016 23:33:38 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Liu",
"Yu",
""
],
[
"Guo",
"Yanming",
""
],
[
"Lew",
"Michael S.",
""
]
] | TITLE: On the Exploration of Convolutional Fusion Networks for Visual
Recognition
ABSTRACT: Despite recent advances in multi-scale deep representations, their
limitations are attributed to expensive parameters and weak fusion modules.
Hence, we propose an efficient approach to fuse multi-scale deep
representations, called convolutional fusion networks (CFN). Owing to using
1$\times$1 convolution and global average pooling, CFN can efficiently generate
the side branches while adding few parameters. In addition, we present a
locally-connected fusion module, which can learn adaptive weights for the side
branches and form a discriminatively fused feature. CFN models trained on the
CIFAR and ImageNet datasets demonstrate remarkable improvements over the plain
CNNs. Furthermore, we generalize CFN to three new tasks, including scene
recognition, fine-grained recognition and image retrieval. Our experiments show
that it can obtain consistent improvements towards the transferring tasks.
| no_new_dataset | 0.947186 |
1611.05521 | Yang Wang | Lin Wu, Yang Wang | Robust Hashing for Multi-View Data: Jointly Learning Low-Rank Kernelized
Similarity Consensus and Hash Functions | Accepted to appear in Image and Vision Computing | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning hash functions/codes for similarity search over multi-view data is
attracting increasing attention, where similar hash codes are assigned to the
data objects characterizing consistently neighborhood relationship across
views. Traditional methods in this category inherently suffer three
limitations: 1) they commonly adopt a two-stage scheme where similarity matrix
is first constructed, followed by a subsequent hash function learning; 2) these
methods are commonly developed on the assumption that data samples with
multiple representations are noise-free,which is not practical in real-life
applications; 3) they often incur cumbersome training model caused by the
neighborhood graph construction using all $N$ points in the database ($O(N)$).
In this paper, we motivate the problem of jointly and efficiently training the
robust hash functions over data objects with multi-feature representations
which may be noise corrupted. To achieve both the robustness and training
efficiency, we propose an approach to effectively and efficiently learning
low-rank kernelized \footnote{We use kernelized similarity rather than kernel,
as it is not a squared symmetric matrix for data-landmark affinity matrix.}
hash functions shared across views. Specifically, we utilize landmark graphs to
construct tractable similarity matrices in multi-views to automatically
discover neighborhood structure in the data. To learn robust hash functions, a
latent low-rank kernel function is used to construct hash functions in order to
accommodate linearly inseparable data. In particular, a latent kernelized
similarity matrix is recovered by rank minimization on multiple kernel-based
similarity matrices. Extensive experiments on real-world multi-view datasets
validate the efficacy of our method in the presence of error corruptions.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 01:21:26 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Wu",
"Lin",
""
],
[
"Wang",
"Yang",
""
]
] | TITLE: Robust Hashing for Multi-View Data: Jointly Learning Low-Rank Kernelized
Similarity Consensus and Hash Functions
ABSTRACT: Learning hash functions/codes for similarity search over multi-view data is
attracting increasing attention, where similar hash codes are assigned to the
data objects characterizing consistently neighborhood relationship across
views. Traditional methods in this category inherently suffer three
limitations: 1) they commonly adopt a two-stage scheme where similarity matrix
is first constructed, followed by a subsequent hash function learning; 2) these
methods are commonly developed on the assumption that data samples with
multiple representations are noise-free,which is not practical in real-life
applications; 3) they often incur cumbersome training model caused by the
neighborhood graph construction using all $N$ points in the database ($O(N)$).
In this paper, we motivate the problem of jointly and efficiently training the
robust hash functions over data objects with multi-feature representations
which may be noise corrupted. To achieve both the robustness and training
efficiency, we propose an approach to effectively and efficiently learning
low-rank kernelized \footnote{We use kernelized similarity rather than kernel,
as it is not a squared symmetric matrix for data-landmark affinity matrix.}
hash functions shared across views. Specifically, we utilize landmark graphs to
construct tractable similarity matrices in multi-views to automatically
discover neighborhood structure in the data. To learn robust hash functions, a
latent low-rank kernel function is used to construct hash functions in order to
accommodate linearly inseparable data. In particular, a latent kernelized
similarity matrix is recovered by rank minimization on multiple kernel-based
similarity matrices. Extensive experiments on real-world multi-view datasets
validate the efficacy of our method in the presence of error corruptions.
| no_new_dataset | 0.952926 |
1611.05592 | Junbo Wang | Junbo Wang, Wei Wang, Yan Huang, Liang Wang, Tieniu Tan | Multimodal Memory Modelling for Video Captioning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Video captioning which automatically translates video clips into natural
language sentences is a very important task in computer vision. By virtue of
recent deep learning technologies, e.g., convolutional neural networks (CNNs)
and recurrent neural networks (RNNs), video captioning has made great progress.
However, learning an effective mapping from visual sequence space to language
space is still a challenging problem. In this paper, we propose a Multimodal
Memory Model (M3) to describe videos, which builds a visual and textual shared
memory to model the long-term visual-textual dependency and further guide
global visual attention on described targets. Specifically, the proposed M3
attaches an external memory to store and retrieve both visual and textual
contents by interacting with video and sentence with multiple read and write
operations. First, text representation in the Long Short-Term Memory (LSTM)
based text decoder is written into the memory, and the memory contents will be
read out to guide an attention to select related visual targets. Then, the
selected visual information is written into the memory, which will be further
read out to the text decoder. To evaluate the proposed model, we perform
experiments on two publicly benchmark datasets: MSVD and MSR-VTT. The
experimental results demonstrate that our method outperforms the
state-of-theart methods in terms of BLEU and METEOR.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 07:24:03 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Wang",
"Junbo",
""
],
[
"Wang",
"Wei",
""
],
[
"Huang",
"Yan",
""
],
[
"Wang",
"Liang",
""
],
[
"Tan",
"Tieniu",
""
]
] | TITLE: Multimodal Memory Modelling for Video Captioning
ABSTRACT: Video captioning which automatically translates video clips into natural
language sentences is a very important task in computer vision. By virtue of
recent deep learning technologies, e.g., convolutional neural networks (CNNs)
and recurrent neural networks (RNNs), video captioning has made great progress.
However, learning an effective mapping from visual sequence space to language
space is still a challenging problem. In this paper, we propose a Multimodal
Memory Model (M3) to describe videos, which builds a visual and textual shared
memory to model the long-term visual-textual dependency and further guide
global visual attention on described targets. Specifically, the proposed M3
attaches an external memory to store and retrieve both visual and textual
contents by interacting with video and sentence with multiple read and write
operations. First, text representation in the Long Short-Term Memory (LSTM)
based text decoder is written into the memory, and the memory contents will be
read out to guide an attention to select related visual targets. Then, the
selected visual information is written into the memory, which will be further
read out to the text decoder. To evaluate the proposed model, we perform
experiments on two publicly benchmark datasets: MSVD and MSR-VTT. The
experimental results demonstrate that our method outperforms the
state-of-theart methods in terms of BLEU and METEOR.
| no_new_dataset | 0.949435 |
1611.05603 | Kai Yu | Kai Yu, Biao Leng, Zhang Zhang, Dangwei Li, Kaiqi Huang | Weakly-supervised Learning of Mid-level Features for Pedestrian
Attribute Recognition and Localization | Containing 9 pages and 5 figures. Codes open-sourced on
https://github.com/kyu-sz/WPAL-network | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | State-of-the-art methods treat pedestrian attribute recognition as a
multi-label image classification problem. The location information of person
attributes is usually eliminated or simply encoded in the rigid splitting of
whole body in previous work. In this paper, we formulate the task in a
weakly-supervised attribute localization framework. Based on GoogLeNet,
firstly, a set of mid-level attribute features are discovered by novelly
designed detection layers, where a max-pooling based weakly-supervised object
detection technique is used to train these layers with only image-level labels
without the need of bounding box annotations of pedestrian attributes.
Secondly, attribute labels are predicted by regression of the detection
response magnitudes. Finally, the locations and rough shapes of pedestrian
attributes can be inferred by performing clustering on a fusion of activation
maps of the detection layers, where the fusion weights are estimated as the
correlation strengths between each attribute and its relevant mid-level
features. Extensive experiments are performed on the two currently largest
pedestrian attribute datasets, i.e. the PETA dataset and the RAP dataset.
Results show that the proposed method has achieved competitive performance on
attribute recognition, compared to other state-of-the-art methods. Moreover,
the results of attribute localization are visualized to understand the
characteristics of the proposed method.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 08:20:23 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Yu",
"Kai",
""
],
[
"Leng",
"Biao",
""
],
[
"Zhang",
"Zhang",
""
],
[
"Li",
"Dangwei",
""
],
[
"Huang",
"Kaiqi",
""
]
] | TITLE: Weakly-supervised Learning of Mid-level Features for Pedestrian
Attribute Recognition and Localization
ABSTRACT: State-of-the-art methods treat pedestrian attribute recognition as a
multi-label image classification problem. The location information of person
attributes is usually eliminated or simply encoded in the rigid splitting of
whole body in previous work. In this paper, we formulate the task in a
weakly-supervised attribute localization framework. Based on GoogLeNet,
firstly, a set of mid-level attribute features are discovered by novelly
designed detection layers, where a max-pooling based weakly-supervised object
detection technique is used to train these layers with only image-level labels
without the need of bounding box annotations of pedestrian attributes.
Secondly, attribute labels are predicted by regression of the detection
response magnitudes. Finally, the locations and rough shapes of pedestrian
attributes can be inferred by performing clustering on a fusion of activation
maps of the detection layers, where the fusion weights are estimated as the
correlation strengths between each attribute and its relevant mid-level
features. Extensive experiments are performed on the two currently largest
pedestrian attribute datasets, i.e. the PETA dataset and the RAP dataset.
Results show that the proposed method has achieved competitive performance on
attribute recognition, compared to other state-of-the-art methods. Moreover,
the results of attribute localization are visualized to understand the
characteristics of the proposed method.
| no_new_dataset | 0.947137 |
1611.05735 | Yaniv Altshuler | Yaniv Altshuler, Alex Pentland, Shlomo Bekhor, Yoram Shiftan, Alfred
Bruckstein | Optimal Dynamic Coverage Infrastructure for Large-Scale Fleets of
Reconnaissance UAVs | 35 pages, 19 figures | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current state of the art in the field of UAV activation relies solely on
human operators for the design and adaptation of the drones' flying routes.
Furthermore, this is being done today on an individual level (one vehicle per
operators), with some exceptions of a handful of new systems, that are
comprised of a small number of self-organizing swarms, manually guided by a
human operator.
Drones-based monitoring is of great importance in variety of civilian
domains, such as road safety, homeland security, and even environmental
control. In its military aspect, efficiently detecting evading targets by a
fleet of unmanned drones has an ever increasing impact on the ability of modern
armies to engage in warfare. The latter is true both traditional symmetric
conflicts among armies as well as asymmetric ones. Be it a speeding driver, a
polluting trailer or a covert convoy, the basic challenge remains the same --
how can its detection probability be maximized using as little number of drones
as possible.
In this work we propose a novel approach for the optimization of large scale
swarms of reconnaissance drones -- capable of producing on-demand optimal
coverage strategies for any given search scenario. Given an estimation cost of
the threat's potential damages, as well as types of monitoring drones available
and their comparative performance, our proposed method generates an
analytically provable strategy, stating the optimal number and types of drones
to be deployed, in order to cost-efficiently monitor a pre-defined region for
targets maneuvering using a given roads networks.
We demonstrate our model using a unique dataset of the Israeli transportation
network, on which different deployment schemes for drones deployment are
evaluated.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 15:28:14 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Altshuler",
"Yaniv",
""
],
[
"Pentland",
"Alex",
""
],
[
"Bekhor",
"Shlomo",
""
],
[
"Shiftan",
"Yoram",
""
],
[
"Bruckstein",
"Alfred",
""
]
] | TITLE: Optimal Dynamic Coverage Infrastructure for Large-Scale Fleets of
Reconnaissance UAVs
ABSTRACT: Current state of the art in the field of UAV activation relies solely on
human operators for the design and adaptation of the drones' flying routes.
Furthermore, this is being done today on an individual level (one vehicle per
operators), with some exceptions of a handful of new systems, that are
comprised of a small number of self-organizing swarms, manually guided by a
human operator.
Drones-based monitoring is of great importance in variety of civilian
domains, such as road safety, homeland security, and even environmental
control. In its military aspect, efficiently detecting evading targets by a
fleet of unmanned drones has an ever increasing impact on the ability of modern
armies to engage in warfare. The latter is true both traditional symmetric
conflicts among armies as well as asymmetric ones. Be it a speeding driver, a
polluting trailer or a covert convoy, the basic challenge remains the same --
how can its detection probability be maximized using as little number of drones
as possible.
In this work we propose a novel approach for the optimization of large scale
swarms of reconnaissance drones -- capable of producing on-demand optimal
coverage strategies for any given search scenario. Given an estimation cost of
the threat's potential damages, as well as types of monitoring drones available
and their comparative performance, our proposed method generates an
analytically provable strategy, stating the optimal number and types of drones
to be deployed, in order to cost-efficiently monitor a pre-defined region for
targets maneuvering using a given roads networks.
We demonstrate our model using a unique dataset of the Israeli transportation
network, on which different deployment schemes for drones deployment are
evaluated.
| new_dataset | 0.851645 |
1611.05751 | Hamid Reza Hassanzadeh | Hamid Reza Hassanzadeh, John H. Phan, May D. Wang | A Multi-Modal Graph-Based Semi-Supervised Pipeline for Predicting Cancer
Survival | in 2016 IEEE International Conference on Bioinformatics and
Biomedicine (BIBM) | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Cancer survival prediction is an active area of research that can help
prevent unnecessary therapies and improve patient's quality of life. Gene
expression profiling is being widely used in cancer studies to discover
informative biomarkers that aid predict different clinical endpoint prediction.
We use multiple modalities of data derived from RNA deep-sequencing (RNA-seq)
to predict survival of cancer patients. Despite the wealth of information
available in expression profiles of cancer tumors, fulfilling the
aforementioned objective remains a big challenge, for the most part, due to the
paucity of data samples compared to the high dimension of the expression
profiles. As such, analysis of transcriptomic data modalities calls for
state-of-the-art big-data analytics techniques that can maximally use all the
available data to discover the relevant information hidden within a significant
amount of noise. In this paper, we propose a pipeline that predicts cancer
patients' survival by exploiting the structure of the input (manifold learning)
and by leveraging the unlabeled samples using Laplacian support vector
machines, a graph-based semi supervised learning (GSSL) paradigm. We show that
under certain circumstances, no single modality per se will result in the best
accuracy and by fusing different models together via a stacked generalization
strategy, we may boost the accuracy synergistically. We apply our approach to
two cancer datasets and present promising results. We maintain that a similar
pipeline can be used for predictive tasks where labeled samples are expensive
to acquire.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 16:01:36 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Hassanzadeh",
"Hamid Reza",
""
],
[
"Phan",
"John H.",
""
],
[
"Wang",
"May D.",
""
]
] | TITLE: A Multi-Modal Graph-Based Semi-Supervised Pipeline for Predicting Cancer
Survival
ABSTRACT: Cancer survival prediction is an active area of research that can help
prevent unnecessary therapies and improve patient's quality of life. Gene
expression profiling is being widely used in cancer studies to discover
informative biomarkers that aid predict different clinical endpoint prediction.
We use multiple modalities of data derived from RNA deep-sequencing (RNA-seq)
to predict survival of cancer patients. Despite the wealth of information
available in expression profiles of cancer tumors, fulfilling the
aforementioned objective remains a big challenge, for the most part, due to the
paucity of data samples compared to the high dimension of the expression
profiles. As such, analysis of transcriptomic data modalities calls for
state-of-the-art big-data analytics techniques that can maximally use all the
available data to discover the relevant information hidden within a significant
amount of noise. In this paper, we propose a pipeline that predicts cancer
patients' survival by exploiting the structure of the input (manifold learning)
and by leveraging the unlabeled samples using Laplacian support vector
machines, a graph-based semi supervised learning (GSSL) paradigm. We show that
under certain circumstances, no single modality per se will result in the best
accuracy and by fusing different models together via a stacked generalization
strategy, we may boost the accuracy synergistically. We apply our approach to
two cancer datasets and present promising results. We maintain that a similar
pipeline can be used for predictive tasks where labeled samples are expensive
to acquire.
| no_new_dataset | 0.949106 |
1611.05755 | Alan Godoy | Guilherme Folego, Marcus A. Angeloni, Jos\'e Augusto Stuchi, Alan
Godoy, Anderson Rocha | Cross-Domain Face Verification: Matching ID Document and Self-Portrait
Photographs | XII WORKSHOP DE VIS\~AO COMPUTACIONAL (Campo Grande, Brazil). In XII
Workshop de Vis\~ao Computacional (pp. 311-316) (2016) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Cross-domain biometrics has been emerging as a new necessity, which poses
several additional challenges, including harsh illumination changes, noise,
pose variation, among others. In this paper, we explore approaches to
cross-domain face verification, comparing self-portrait photographs ("selfies")
to ID documents. We approach the problem with proper image photometric
adjustment and data standardization techniques, along with deep learning
methods to extract the most prominent features from the data, reducing the
effects of domain shift in this problem. We validate the methods using a novel
dataset comprising 50 individuals. The obtained results are promising and
indicate that the adopted path is worth further investigation.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 16:05:11 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Folego",
"Guilherme",
""
],
[
"Angeloni",
"Marcus A.",
""
],
[
"Stuchi",
"José Augusto",
""
],
[
"Godoy",
"Alan",
""
],
[
"Rocha",
"Anderson",
""
]
] | TITLE: Cross-Domain Face Verification: Matching ID Document and Self-Portrait
Photographs
ABSTRACT: Cross-domain biometrics has been emerging as a new necessity, which poses
several additional challenges, including harsh illumination changes, noise,
pose variation, among others. In this paper, we explore approaches to
cross-domain face verification, comparing self-portrait photographs ("selfies")
to ID documents. We approach the problem with proper image photometric
adjustment and data standardization techniques, along with deep learning
methods to extract the most prominent features from the data, reducing the
effects of domain shift in this problem. We validate the methods using a novel
dataset comprising 50 individuals. The obtained results are promising and
indicate that the adopted path is worth further investigation.
| new_dataset | 0.956145 |
1611.05777 | Hamid Reza Hassanzadeh | Hamid Reza Hassanzadeh, May D. Wang | DeeperBind: Enhancing Prediction of Sequence Specificities of DNA
Binding Proteins | in 2016 IEEE International Conference on Bioinformatics and
Biomedicine (BIBM) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Transcription factors (TFs) are macromolecules that bind to
\textit{cis}-regulatory specific sub-regions of DNA promoters and initiate
transcription. Finding the exact location of these binding sites (aka motifs)
is important in a variety of domains such as drug design and development. To
address this need, several \textit{in vivo} and \textit{in vitro} techniques
have been developed so far that try to characterize and predict the binding
specificity of a protein to different DNA loci. The major problem with these
techniques is that they are not accurate enough in prediction of the binding
affinity and characterization of the corresponding motifs. As a result,
downstream analysis is required to uncover the locations where proteins of
interest bind. Here, we propose DeeperBind, a long short term recurrent
convolutional network for prediction of protein binding specificities with
respect to DNA probes. DeeperBind can model the positional dynamics of probe
sequences and hence reckons with the contributions made by individual
sub-regions in DNA sequences, in an effective way. Moreover, it can be trained
and tested on datasets containing varying-length sequences. We apply our
pipeline to the datasets derived from protein binding microarrays (PBMs), an
in-vitro high-throughput technology for quantification of protein-DNA binding
preferences, and present promising results. To the best of our knowledge, this
is the most accurate pipeline that can predict binding specificities of DNA
sequences from the data produced by high-throughput technologies through
utilization of the power of deep learning for feature generation and positional
dynamics modeling.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 16:52:41 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Hassanzadeh",
"Hamid Reza",
""
],
[
"Wang",
"May D.",
""
]
] | TITLE: DeeperBind: Enhancing Prediction of Sequence Specificities of DNA
Binding Proteins
ABSTRACT: Transcription factors (TFs) are macromolecules that bind to
\textit{cis}-regulatory specific sub-regions of DNA promoters and initiate
transcription. Finding the exact location of these binding sites (aka motifs)
is important in a variety of domains such as drug design and development. To
address this need, several \textit{in vivo} and \textit{in vitro} techniques
have been developed so far that try to characterize and predict the binding
specificity of a protein to different DNA loci. The major problem with these
techniques is that they are not accurate enough in prediction of the binding
affinity and characterization of the corresponding motifs. As a result,
downstream analysis is required to uncover the locations where proteins of
interest bind. Here, we propose DeeperBind, a long short term recurrent
convolutional network for prediction of protein binding specificities with
respect to DNA probes. DeeperBind can model the positional dynamics of probe
sequences and hence reckons with the contributions made by individual
sub-regions in DNA sequences, in an effective way. Moreover, it can be trained
and tested on datasets containing varying-length sequences. We apply our
pipeline to the datasets derived from protein binding microarrays (PBMs), an
in-vitro high-throughput technology for quantification of protein-DNA binding
preferences, and present promising results. To the best of our knowledge, this
is the most accurate pipeline that can predict binding specificities of DNA
sequences from the data produced by high-throughput technologies through
utilization of the power of deep learning for feature generation and positional
dynamics modeling.
| no_new_dataset | 0.949856 |
1611.05799 | Nichola Abdo | Philipp Jund, Nichola Abdo, Andreas Eitel, Wolfram Burgard | The Freiburg Groceries Dataset | Link to dataset:
http://www2.informatik.uni-freiburg.de/~eitel/freiburg_groceries_dataset.html
Link to code: https://github.com/PhilJd/freiburg_groceries_dataset | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the increasing performance of machine learning techniques in the last
few years, the computer vision and robotics communities have created a large
number of datasets for benchmarking object recognition tasks. These datasets
cover a large spectrum of natural images and object categories, making them not
only useful as a testbed for comparing machine learning approaches, but also a
great resource for bootstrapping different domain-specific perception and
robotic systems. One such domain is domestic environments, where an autonomous
robot has to recognize a large variety of everyday objects such as groceries.
This is a challenging task due to the large variety of objects and products,
and where there is great need for real-world training data that goes beyond
product images available online. In this paper, we address this issue and
present a dataset consisting of 5,000 images covering 25 different classes of
groceries, with at least 97 images per class. We collected all images from
real-world settings at different stores and apartments. In contrast to existing
groceries datasets, our dataset includes a large variety of perspectives,
lighting conditions, and degrees of clutter. Overall, our images contain
thousands of different object instances. It is our hope that machine learning
and robotics researchers find this dataset of use for training, testing, and
bootstrapping their approaches. As a baseline classifier to facilitate
comparison, we re-trained the CaffeNet architecture (an adaptation of the
well-known AlexNet) on our dataset and achieved a mean accuracy of 78.9%. We
release this trained model along with the code and data splits we used in our
experiments.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 17:35:21 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Jund",
"Philipp",
""
],
[
"Abdo",
"Nichola",
""
],
[
"Eitel",
"Andreas",
""
],
[
"Burgard",
"Wolfram",
""
]
] | TITLE: The Freiburg Groceries Dataset
ABSTRACT: With the increasing performance of machine learning techniques in the last
few years, the computer vision and robotics communities have created a large
number of datasets for benchmarking object recognition tasks. These datasets
cover a large spectrum of natural images and object categories, making them not
only useful as a testbed for comparing machine learning approaches, but also a
great resource for bootstrapping different domain-specific perception and
robotic systems. One such domain is domestic environments, where an autonomous
robot has to recognize a large variety of everyday objects such as groceries.
This is a challenging task due to the large variety of objects and products,
and where there is great need for real-world training data that goes beyond
product images available online. In this paper, we address this issue and
present a dataset consisting of 5,000 images covering 25 different classes of
groceries, with at least 97 images per class. We collected all images from
real-world settings at different stores and apartments. In contrast to existing
groceries datasets, our dataset includes a large variety of perspectives,
lighting conditions, and degrees of clutter. Overall, our images contain
thousands of different object instances. It is our hope that machine learning
and robotics researchers find this dataset of use for training, testing, and
bootstrapping their approaches. As a baseline classifier to facilitate
comparison, we re-trained the CaffeNet architecture (an adaptation of the
well-known AlexNet) on our dataset and achieved a mean accuracy of 78.9%. We
release this trained model along with the code and data splits we used in our
experiments.
| new_dataset | 0.967287 |
1611.05803 | Mohamed Nait Meziane | Thomas Picon, Mohamed Nait Meziane, Philippe Ravier, Guy Lamarque,
Clarisse Novello, Jean-Charles Le Bunetel, Yves Raingeaud | COOLL: Controlled On/Off Loads Library, a Public Dataset of High-Sampled
Electrical Signals for Appliance Identification | 5 pages, 2 figures, 3 tables | null | null | null | cs.OH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper gives a brief description of the Controlled On/Off Loads Library
(COOLL) dataset. This latter is a dataset of high-sampled electrical current
and voltage measurements representing individual appliances consumption. The
measurements were taken in June 2016 in the PRISME laboratory of the University
of Orl\'eans, France. The appliances are mainly controllable appliances (i.e.
we can precisely control their turn-on/off time instants). 42 appliances of 12
types were measured at a 100 kHz sampling frequency.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 18:03:05 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Picon",
"Thomas",
""
],
[
"Meziane",
"Mohamed Nait",
""
],
[
"Ravier",
"Philippe",
""
],
[
"Lamarque",
"Guy",
""
],
[
"Novello",
"Clarisse",
""
],
[
"Bunetel",
"Jean-Charles Le",
""
],
[
"Raingeaud",
"Yves",
""
]
] | TITLE: COOLL: Controlled On/Off Loads Library, a Public Dataset of High-Sampled
Electrical Signals for Appliance Identification
ABSTRACT: This paper gives a brief description of the Controlled On/Off Loads Library
(COOLL) dataset. This latter is a dataset of high-sampled electrical current
and voltage measurements representing individual appliances consumption. The
measurements were taken in June 2016 in the PRISME laboratory of the University
of Orl\'eans, France. The appliances are mainly controllable appliances (i.e.
we can precisely control their turn-on/off time instants). 42 appliances of 12
types were measured at a 100 kHz sampling frequency.
| new_dataset | 0.937555 |
1611.05842 | Dilip K. Prasad | D. K. Prasad, D. Rajan, L. Rachmawati, E. Rajabaly, C. Quek | Video Processing from Electro-optical Sensors for Object Detection and
Tracking in Maritime Environment: A Survey | 23 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a survey on maritime object detection and tracking approaches,
which are essential for the development of a navigational system for autonomous
ships. The electro-optical (EO) sensor considered here is a video camera that
operates in the visible or the infrared spectra, which conventionally
complement radar and sonar and have demonstrated effectiveness for situational
awareness at sea has demonstrated its effectiveness over the last few years.
This paper provides a comprehensive overview of various approaches of video
processing for object detection and tracking in the maritime environment. We
follow an approach-based taxonomy wherein the advantages and limitations of
each approach are compared. The object detection system consists of the
following modules: horizon detection, static background subtraction and
foreground segmentation. Each of these has been studied extensively in maritime
situations and has been shown to be challenging due to the presence of
background motion especially due to waves and wakes. The main processes
involved in object tracking include video frame registration, dynamic
background subtraction, and the object tracking algorithm itself. The
challenges for robust tracking arise due to camera motion, dynamic background
and low contrast of tracked object, possibly due to environmental degradation.
The survey also discusses multisensor approaches and commercial maritime
systems that use EO sensors. The survey also highlights methods from computer
vision research which hold promise to perform well in maritime EO data
processing. Performance of several maritime and computer vision techniques is
evaluated on newly proposed Singapore Maritime Dataset.
| [
{
"version": "v1",
"created": "Thu, 17 Nov 2016 20:11:51 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Prasad",
"D. K.",
""
],
[
"Rajan",
"D.",
""
],
[
"Rachmawati",
"L.",
""
],
[
"Rajabaly",
"E.",
""
],
[
"Quek",
"C.",
""
]
] | TITLE: Video Processing from Electro-optical Sensors for Object Detection and
Tracking in Maritime Environment: A Survey
ABSTRACT: We present a survey on maritime object detection and tracking approaches,
which are essential for the development of a navigational system for autonomous
ships. The electro-optical (EO) sensor considered here is a video camera that
operates in the visible or the infrared spectra, which conventionally
complement radar and sonar and have demonstrated effectiveness for situational
awareness at sea has demonstrated its effectiveness over the last few years.
This paper provides a comprehensive overview of various approaches of video
processing for object detection and tracking in the maritime environment. We
follow an approach-based taxonomy wherein the advantages and limitations of
each approach are compared. The object detection system consists of the
following modules: horizon detection, static background subtraction and
foreground segmentation. Each of these has been studied extensively in maritime
situations and has been shown to be challenging due to the presence of
background motion especially due to waves and wakes. The main processes
involved in object tracking include video frame registration, dynamic
background subtraction, and the object tracking algorithm itself. The
challenges for robust tracking arise due to camera motion, dynamic background
and low contrast of tracked object, possibly due to environmental degradation.
The survey also discusses multisensor approaches and commercial maritime
systems that use EO sensors. The survey also highlights methods from computer
vision research which hold promise to perform well in maritime EO data
processing. Performance of several maritime and computer vision techniques is
evaluated on newly proposed Singapore Maritime Dataset.
| no_new_dataset | 0.948822 |
physics/0701084 | Alessandra Retico | P. Delogu, M.E. Fantacci, A. Preite Martinez, A. Retico, A. Stefanini,
A. Tata | A scalable system for microcalcification cluster automated detection in
a distributed mammographic database | 6 pages, 4 figures; Proceedings of the IEEE NNS and MIC Conference,
October 23-29, 2005, Puerto Rico | Nuclear Science Symposium Conference Record, 2005 IEEE, Volume 3,
23-29 Oct. 2005, 5 pp | 10.1109/NSSMIC.2005.1596609 | null | physics.med-ph | null | A computer-aided detection (CADe) system for microcalcification cluster
identification in mammograms has been developed in the framework of the
EU-founded MammoGrid project. The CADe software is mainly based on wavelet
transforms and artificial neural networks. It is able to identify
microcalcifications in different datasets of mammograms (i.e. acquired with
different machines and settings, digitized with different pitch and bit depth
or direct digital ones). The CADe can be remotely run from GRID-connected
acquisition and annotation stations, supporting clinicians from geographically
distant locations in the interpretation of mammographic data. We report and
discuss the system performances on different datasets of mammograms and the
status of the GRID-enabled CADe analysis.
| [
{
"version": "v1",
"created": "Mon, 8 Jan 2007 11:08:52 GMT"
}
] | 2016-11-18T00:00:00 | [
[
"Delogu",
"P.",
""
],
[
"Fantacci",
"M. E.",
""
],
[
"Martinez",
"A. Preite",
""
],
[
"Retico",
"A.",
""
],
[
"Stefanini",
"A.",
""
],
[
"Tata",
"A.",
""
]
] | TITLE: A scalable system for microcalcification cluster automated detection in
a distributed mammographic database
ABSTRACT: A computer-aided detection (CADe) system for microcalcification cluster
identification in mammograms has been developed in the framework of the
EU-founded MammoGrid project. The CADe software is mainly based on wavelet
transforms and artificial neural networks. It is able to identify
microcalcifications in different datasets of mammograms (i.e. acquired with
different machines and settings, digitized with different pitch and bit depth
or direct digital ones). The CADe can be remotely run from GRID-connected
acquisition and annotation stations, supporting clinicians from geographically
distant locations in the interpretation of mammographic data. We report and
discuss the system performances on different datasets of mammograms and the
status of the GRID-enabled CADe analysis.
| no_new_dataset | 0.953362 |
0808.3546 | Ioan Raicu | Ioan Raicu, Yong Zhao, Ian Foster, Alex Szalay | Accelerating Large-scale Data Exploration through Data Diffusion | IEEE/ACM International Workshop on Data-Aware Distributed Computing
2008 | null | 10.1145/1383519.1383521 | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data-intensive applications often require exploratory analysis of large
datasets. If analysis is performed on distributed resources, data locality can
be crucial to high throughput and performance. We propose a "data diffusion"
approach that acquires compute and storage resources dynamically, replicates
data in response to demand, and schedules computations close to data. As demand
increases, more resources are acquired, thus allowing faster response to
subsequent requests that refer to the same data; when demand drops, resources
are released. This approach can provide the benefits of dedicated hardware
without the associated high costs, depending on workload and resource
characteristics. The approach is reminiscent of cooperative caching,
web-caching, and peer-to-peer storage systems, but addresses different
application demands. Other data-aware scheduling approaches assume dedicated
resources, which can be expensive and/or inefficient if load varies
significantly. To explore the feasibility of the data diffusion approach, we
have extended the Falkon resource provisioning and task scheduling system to
support data caching and data-aware scheduling. Performance results from both
micro-benchmarks and a large scale astronomy application demonstrate that our
approach improves performance relative to alternative approaches, as well as
provides improved scalability as aggregated I/O bandwidth scales linearly with
the number of data cache nodes.
| [
{
"version": "v1",
"created": "Tue, 26 Aug 2008 16:02:50 GMT"
}
] | 2016-11-17T00:00:00 | [
[
"Raicu",
"Ioan",
""
],
[
"Zhao",
"Yong",
""
],
[
"Foster",
"Ian",
""
],
[
"Szalay",
"Alex",
""
]
] | TITLE: Accelerating Large-scale Data Exploration through Data Diffusion
ABSTRACT: Data-intensive applications often require exploratory analysis of large
datasets. If analysis is performed on distributed resources, data locality can
be crucial to high throughput and performance. We propose a "data diffusion"
approach that acquires compute and storage resources dynamically, replicates
data in response to demand, and schedules computations close to data. As demand
increases, more resources are acquired, thus allowing faster response to
subsequent requests that refer to the same data; when demand drops, resources
are released. This approach can provide the benefits of dedicated hardware
without the associated high costs, depending on workload and resource
characteristics. The approach is reminiscent of cooperative caching,
web-caching, and peer-to-peer storage systems, but addresses different
application demands. Other data-aware scheduling approaches assume dedicated
resources, which can be expensive and/or inefficient if load varies
significantly. To explore the feasibility of the data diffusion approach, we
have extended the Falkon resource provisioning and task scheduling system to
support data caching and data-aware scheduling. Performance results from both
micro-benchmarks and a large scale astronomy application demonstrate that our
approach improves performance relative to alternative approaches, as well as
provides improved scalability as aggregated I/O bandwidth scales linearly with
the number of data cache nodes.
| no_new_dataset | 0.947478 |
0904.3310 | Shariq Bashir Mr. | Shariq Bashir, Abdul Rauf Baig | FastLMFI: An Efficient Approach for Local Maximal Patterns Propagation
and Maximal Patterns Superset Checking | 8 Pages, In the proceedings of 4th ACS/IEEE International Conference
on Computer Systems and Applications 2006, March 8, 2006, Dubai/Sharjah, UAE,
2006, Page(s) 452-459 | null | 10.1109/AICCSA.2006.205130 | null | cs.DB cs.AI cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Maximal frequent patterns superset checking plays an important role in the
efficient mining of complete Maximal Frequent Itemsets (MFI) and maximal search
space pruning. In this paper we present a new indexing approach, FastLMFI for
local maximal frequent patterns (itemset) propagation and maximal patterns
superset checking. Experimental results on different sparse and dense datasets
show that our work is better than the previous well known progressive focusing
technique. We have also integrated our superset checking approach with an
existing state of the art maximal itemsets algorithm Mafia, and compare our
results with current best maximal itemsets algorithms afopt-max and FP
(zhu)-max. Our results outperform afopt-max and FP (zhu)-max on dense (chess
and mushroom) datasets on almost all support thresholds, which shows the
effectiveness of our approach.
| [
{
"version": "v1",
"created": "Tue, 21 Apr 2009 18:33:04 GMT"
}
] | 2016-11-17T00:00:00 | [
[
"Bashir",
"Shariq",
""
],
[
"Baig",
"Abdul Rauf",
""
]
] | TITLE: FastLMFI: An Efficient Approach for Local Maximal Patterns Propagation
and Maximal Patterns Superset Checking
ABSTRACT: Maximal frequent patterns superset checking plays an important role in the
efficient mining of complete Maximal Frequent Itemsets (MFI) and maximal search
space pruning. In this paper we present a new indexing approach, FastLMFI for
local maximal frequent patterns (itemset) propagation and maximal patterns
superset checking. Experimental results on different sparse and dense datasets
show that our work is better than the previous well known progressive focusing
technique. We have also integrated our superset checking approach with an
existing state of the art maximal itemsets algorithm Mafia, and compare our
results with current best maximal itemsets algorithms afopt-max and FP
(zhu)-max. Our results outperform afopt-max and FP (zhu)-max on dense (chess
and mushroom) datasets on almost all support thresholds, which shows the
effectiveness of our approach.
| no_new_dataset | 0.950549 |
0904.3312 | Shariq Bashir Mr. | Shariq Bashir, and Abdul Rauf Baig | HybridMiner: Mining Maximal Frequent Itemsets Using Hybrid Database
Representation Approach | 8 Pages In the proceedings of 9th IEEE-INMIC 2005, Karachi, Pakistan,
2005 | null | 10.1109/INMIC.2005.334484 | null | cs.DB cs.AI cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we present a novel hybrid (arraybased layout and vertical
bitmap layout) database representation approach for mining complete Maximal
Frequent Itemset (MFI) on sparse and large datasets. Our work is novel in terms
of scalability, item search order and two horizontal and vertical projection
techniques. We also present a maximal algorithm using this hybrid database
representation approach. Different experimental results on real and sparse
benchmark datasets show that our approach is better than previous state of art
maximal algorithms.
| [
{
"version": "v1",
"created": "Tue, 21 Apr 2009 18:38:25 GMT"
}
] | 2016-11-17T00:00:00 | [
[
"Bashir",
"Shariq",
""
],
[
"Baig",
"Abdul Rauf",
""
]
] | TITLE: HybridMiner: Mining Maximal Frequent Itemsets Using Hybrid Database
Representation Approach
ABSTRACT: In this paper we present a novel hybrid (arraybased layout and vertical
bitmap layout) database representation approach for mining complete Maximal
Frequent Itemset (MFI) on sparse and large datasets. Our work is novel in terms
of scalability, item search order and two horizontal and vertical projection
techniques. We also present a maximal algorithm using this hybrid database
representation approach. Different experimental results on real and sparse
benchmark datasets show that our approach is better than previous state of art
maximal algorithms.
| no_new_dataset | 0.948775 |
0906.0391 | Ilya Volnyansky | Ilya Volnyansky, Vladimir Pestov | Curse of Dimensionality in Pivot-based Indexes | 9 pp., 4 figures, latex 2e, a revised submission to the 2nd
International Workshop on Similarity Search and Applications, 2009 | Proc. 2nd Int. Workshop on Similarity Search and Applications
(SISAP 2009), Prague, Aug. 29-30, 2009, T. Skopal and P. Zezula (eds.), IEEE
Computer Society, Los Alamitos--Washington--Tokyo, 2009, pp. 39-46. | 10.1109/SISAP.2009.9 | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We offer a theoretical validation of the curse of dimensionality in the
pivot-based indexing of datasets for similarity search, by proving, in the
framework of statistical learning, that in high dimensions no pivot-based
indexing scheme can essentially outperform the linear scan.
A study of the asymptotic performance of pivot-based indexing schemes is
performed on a sequence of datasets modeled as samples $X_d$ picked in i.i.d.
fashion from metric spaces $\Omega_d$. We allow the size of the dataset $n=n_d$
to be such that $d$, the ``dimension'', is superlogarithmic but subpolynomial
in $n$. The number of pivots is allowed to grow as $o(n/d)$. We pick the least
restrictive cost model of similarity search where we count each distance
calculation as a single computation and disregard the rest.
We demonstrate that if the intrinsic dimension of the spaces $\Omega_d$ in
the sense of concentration of measure phenomenon is $O(d)$, then the
performance of similarity search pivot-based indexes is asymptotically linear
in $n$.
| [
{
"version": "v1",
"created": "Tue, 2 Jun 2009 00:41:46 GMT"
},
{
"version": "v2",
"created": "Thu, 18 Jun 2009 23:50:44 GMT"
}
] | 2016-11-17T00:00:00 | [
[
"Volnyansky",
"Ilya",
""
],
[
"Pestov",
"Vladimir",
""
]
] | TITLE: Curse of Dimensionality in Pivot-based Indexes
ABSTRACT: We offer a theoretical validation of the curse of dimensionality in the
pivot-based indexing of datasets for similarity search, by proving, in the
framework of statistical learning, that in high dimensions no pivot-based
indexing scheme can essentially outperform the linear scan.
A study of the asymptotic performance of pivot-based indexing schemes is
performed on a sequence of datasets modeled as samples $X_d$ picked in i.i.d.
fashion from metric spaces $\Omega_d$. We allow the size of the dataset $n=n_d$
to be such that $d$, the ``dimension'', is superlogarithmic but subpolynomial
in $n$. The number of pivots is allowed to grow as $o(n/d)$. We pick the least
restrictive cost model of similarity search where we count each distance
calculation as a single computation and disregard the rest.
We demonstrate that if the intrinsic dimension of the spaces $\Omega_d$ in
the sense of concentration of measure phenomenon is $O(d)$, then the
performance of similarity search pivot-based indexes is asymptotically linear
in $n$.
| no_new_dataset | 0.943086 |
0911.2904 | Maxim Raginsky | Maxim Raginsky, Rebecca Willett, Corinne Horn, Jorge Silva, Roummel
Marcia | Sequential anomaly detection in the presence of noise and limited
feedback | 19 pages, 12 pdf figures; final version to be published in IEEE
Transactions on Information Theory | null | 10.1109/TIT.2012.2201375 | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper describes a methodology for detecting anomalies from sequentially
observed and potentially noisy data. The proposed approach consists of two main
elements: (1) {\em filtering}, or assigning a belief or likelihood to each
successive measurement based upon our ability to predict it from previous noisy
observations, and (2) {\em hedging}, or flagging potential anomalies by
comparing the current belief against a time-varying and data-adaptive
threshold. The threshold is adjusted based on the available feedback from an
end user. Our algorithms, which combine universal prediction with recent work
on online convex programming, do not require computing posterior distributions
given all current observations and involve simple primal-dual parameter
updates. At the heart of the proposed approach lie exponential-family models
which can be used in a wide variety of contexts and applications, and which
yield methods that achieve sublinear per-round regret against both static and
slowly varying product distributions with marginals drawn from the same
exponential family. Moreover, the regret against static distributions coincides
with the minimax value of the corresponding online strongly convex game. We
also prove bounds on the number of mistakes made during the hedging step
relative to the best offline choice of the threshold with access to all
estimated beliefs and feedback signals. We validate the theory on synthetic
data drawn from a time-varying distribution over binary vectors of high
dimensionality, as well as on the Enron email dataset.
| [
{
"version": "v1",
"created": "Sun, 15 Nov 2009 18:43:10 GMT"
},
{
"version": "v2",
"created": "Wed, 14 Jul 2010 17:30:25 GMT"
},
{
"version": "v3",
"created": "Sun, 5 Feb 2012 23:11:54 GMT"
},
{
"version": "v4",
"created": "Tue, 13 Mar 2012 16:11:21 GMT"
}
] | 2016-11-17T00:00:00 | [
[
"Raginsky",
"Maxim",
""
],
[
"Willett",
"Rebecca",
""
],
[
"Horn",
"Corinne",
""
],
[
"Silva",
"Jorge",
""
],
[
"Marcia",
"Roummel",
""
]
] | TITLE: Sequential anomaly detection in the presence of noise and limited
feedback
ABSTRACT: This paper describes a methodology for detecting anomalies from sequentially
observed and potentially noisy data. The proposed approach consists of two main
elements: (1) {\em filtering}, or assigning a belief or likelihood to each
successive measurement based upon our ability to predict it from previous noisy
observations, and (2) {\em hedging}, or flagging potential anomalies by
comparing the current belief against a time-varying and data-adaptive
threshold. The threshold is adjusted based on the available feedback from an
end user. Our algorithms, which combine universal prediction with recent work
on online convex programming, do not require computing posterior distributions
given all current observations and involve simple primal-dual parameter
updates. At the heart of the proposed approach lie exponential-family models
which can be used in a wide variety of contexts and applications, and which
yield methods that achieve sublinear per-round regret against both static and
slowly varying product distributions with marginals drawn from the same
exponential family. Moreover, the regret against static distributions coincides
with the minimax value of the corresponding online strongly convex game. We
also prove bounds on the number of mistakes made during the hedging step
relative to the best offline choice of the threshold with access to all
estimated beliefs and feedback signals. We validate the theory on synthetic
data drawn from a time-varying distribution over binary vectors of high
dimensionality, as well as on the Enron email dataset.
| no_new_dataset | 0.944638 |
1001.2411 | Uwe Aickelin | Julie Greensmith, Jamie Twycross, Uwe Aickelin | Dendritic Cells for Anomaly Detection | 8 pages, 10 tables, 4 figures, IEEE Congress on Evolutionary
Computation (CEC2006), Vancouver, Canada | Proceedings of the IEEE Congress on Evolutionary Computation
(CEC2006), Vancouver, Canada | 10.1109/CEC.2006.1688374 | null | cs.AI cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Artificial immune systems, more specifically the negative selection
algorithm, have previously been applied to intrusion detection. The aim of this
research is to develop an intrusion detection system based on a novel concept
in immunology, the Danger Theory. Dendritic Cells (DCs) are antigen presenting
cells and key to the activation of the human signals from the host tissue and
correlate these signals with proteins know as antigens. In algorithmic terms,
individual DCs perform multi-sensor data fusion based on time-windows. The
whole population of DCs asynchronously correlates the fused signals with a
secondary data stream. The behaviour of human DCs is abstracted to form the DC
Algorithm (DCA), which is implemented using an immune inspired framework,
libtissue. This system is used to detect context switching for a basic machine
learning dataset and to detect outgoing portscans in real-time. Experimental
results show a significant difference between an outgoing portscan and normal
traffic.
| [
{
"version": "v1",
"created": "Thu, 14 Jan 2010 10:51:41 GMT"
}
] | 2016-11-17T00:00:00 | [
[
"Greensmith",
"Julie",
""
],
[
"Twycross",
"Jamie",
""
],
[
"Aickelin",
"Uwe",
""
]
] | TITLE: Dendritic Cells for Anomaly Detection
ABSTRACT: Artificial immune systems, more specifically the negative selection
algorithm, have previously been applied to intrusion detection. The aim of this
research is to develop an intrusion detection system based on a novel concept
in immunology, the Danger Theory. Dendritic Cells (DCs) are antigen presenting
cells and key to the activation of the human signals from the host tissue and
correlate these signals with proteins know as antigens. In algorithmic terms,
individual DCs perform multi-sensor data fusion based on time-windows. The
whole population of DCs asynchronously correlates the fused signals with a
secondary data stream. The behaviour of human DCs is abstracted to form the DC
Algorithm (DCA), which is implemented using an immune inspired framework,
libtissue. This system is used to detect context switching for a basic machine
learning dataset and to detect outgoing portscans in real-time. Experimental
results show a significant difference between an outgoing portscan and normal
traffic.
| no_new_dataset | 0.943556 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.