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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1601.05273 | Alvin Lebeck | Yang Liu, Chris Dwyer, Alvin R. Lebeck | Combined Compute and Storage: Configurable Memristor Arrays to
Accelerate Search | null | null | null | null | cs.ET | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Emerging technologies present opportunities for system designers to meet the
challenges presented by competing trends of big data analytics and limitations
on CMOS scaling. Specifically, memristors are an emerging high-density
technology where the individual memristors can be used as storage or to perform
computation. The voltage applied across a memristor determines its behavior
(storage vs. compute), which enables a configurable memristor substrate that
can embed computation with storage.
This paper explores accelerating point and range search queries as instances
of the more general configurable combined compute and storage capabilities of
memristor arrays. We first present MemCAM, a configurable memristor-based
content addressable memory for the cases when fast, infrequent searches over
large datasets are required. For frequent searches, memristor lifetime becomes
a concern. To increase memristor array lifetime we introduce hybrid data
structures that combine trees with MemCAM using conventional CMOS
processor/cache hierarchies for the upper levels of the tree and configurable
memristor technologies for lower levels.
We use SPICE to analyze energy consumption and access time of memristors and
use analytic models to evaluate the performance of configurable hybrid data
structures. The results show that with acceptable energy consumption our
configurable hybrid data structures improve performance of search intensive
applications and achieve lifetime in years or decades under continuous queries.
Furthermore, the configurability of memristor arrays and the proposed data
structures provide opportunities to tune the trade- off between performance and
lifetime and the data structures can be easily adapted to future memristors or
other technologies with improved endurance.
| [
{
"version": "v1",
"created": "Wed, 20 Jan 2016 14:08:29 GMT"
}
] | 2016-01-21T00:00:00 | [
[
"Liu",
"Yang",
""
],
[
"Dwyer",
"Chris",
""
],
[
"Lebeck",
"Alvin R.",
""
]
] | TITLE: Combined Compute and Storage: Configurable Memristor Arrays to
Accelerate Search
ABSTRACT: Emerging technologies present opportunities for system designers to meet the
challenges presented by competing trends of big data analytics and limitations
on CMOS scaling. Specifically, memristors are an emerging high-density
technology where the individual memristors can be used as storage or to perform
computation. The voltage applied across a memristor determines its behavior
(storage vs. compute), which enables a configurable memristor substrate that
can embed computation with storage.
This paper explores accelerating point and range search queries as instances
of the more general configurable combined compute and storage capabilities of
memristor arrays. We first present MemCAM, a configurable memristor-based
content addressable memory for the cases when fast, infrequent searches over
large datasets are required. For frequent searches, memristor lifetime becomes
a concern. To increase memristor array lifetime we introduce hybrid data
structures that combine trees with MemCAM using conventional CMOS
processor/cache hierarchies for the upper levels of the tree and configurable
memristor technologies for lower levels.
We use SPICE to analyze energy consumption and access time of memristors and
use analytic models to evaluate the performance of configurable hybrid data
structures. The results show that with acceptable energy consumption our
configurable hybrid data structures improve performance of search intensive
applications and achieve lifetime in years or decades under continuous queries.
Furthermore, the configurability of memristor arrays and the proposed data
structures provide opportunities to tune the trade- off between performance and
lifetime and the data structures can be easily adapted to future memristors or
other technologies with improved endurance.
| no_new_dataset | 0.950869 |
1601.05403 | Jo\~ao Sedoc | Jo\~ao Sedoc, Jean Gallier, Lyle Ungar, Dean Foster | Semantic Word Clusters Using Signed Normalized Graph Cuts | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vector space representations of words capture many aspects of word
similarity, but such methods tend to make vector spaces in which antonyms (as
well as synonyms) are close to each other. We present a new signed spectral
normalized graph cut algorithm, signed clustering, that overlays existing
thesauri upon distributionally derived vector representations of words, so that
antonym relationships between word pairs are represented by negative weights.
Our signed clustering algorithm produces clusters of words which simultaneously
capture distributional and synonym relations. We evaluate these clusters
against the SimLex-999 dataset (Hill et al.,2014) of human judgments of word
pair similarities, and also show the benefit of using our clusters to predict
the sentiment of a given text.
| [
{
"version": "v1",
"created": "Wed, 20 Jan 2016 20:37:47 GMT"
}
] | 2016-01-21T00:00:00 | [
[
"Sedoc",
"João",
""
],
[
"Gallier",
"Jean",
""
],
[
"Ungar",
"Lyle",
""
],
[
"Foster",
"Dean",
""
]
] | TITLE: Semantic Word Clusters Using Signed Normalized Graph Cuts
ABSTRACT: Vector space representations of words capture many aspects of word
similarity, but such methods tend to make vector spaces in which antonyms (as
well as synonyms) are close to each other. We present a new signed spectral
normalized graph cut algorithm, signed clustering, that overlays existing
thesauri upon distributionally derived vector representations of words, so that
antonym relationships between word pairs are represented by negative weights.
Our signed clustering algorithm produces clusters of words which simultaneously
capture distributional and synonym relations. We evaluate these clusters
against the SimLex-999 dataset (Hill et al.,2014) of human judgments of word
pair similarities, and also show the benefit of using our clusters to predict
the sentiment of a given text.
| no_new_dataset | 0.9462 |
1601.05409 | Mitra Montazeri | Mitra Montazeri, Mahdieh Soleymani Baghshah, Aliakbar Niknafs | Selecting Efficient Features via a Hyper-Heuristic Approach | The Fifth Iran Data Mining Conference (IDMC 2011), Amirkabir
University of Technology, Tehran, Iran | null | null | null | cs.CV cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | By Emerging huge databases and the need to efficient learning algorithms on
these datasets, new problems have appeared and some methods have been proposed
to solve these problems by selecting efficient features. Feature selection is a
problem of finding efficient features among all features in which the final
feature set can improve accuracy and reduce complexity. One way to solve this
problem is to evaluate all possible feature subsets. However, evaluating all
possible feature subsets is an exhaustive search and thus it has high
computational complexity. Until now many heuristic algorithms have been studied
for solving this problem. Hyper-heuristic is a new heuristic approach which can
search the solution space effectively by applying local searches appropriately.
Each local search is a neighborhood searching algorithm. Since each region of
the solution space can have its own characteristics, it should be chosen an
appropriate local search and apply it to current solution. This task is tackled
to a supervisor. The supervisor chooses a local search based on the functional
history of local searches. By doing this task, it can trade of between
exploitation and exploration. Since the existing heuristic cannot trade of
between exploration and exploitation appropriately, the solution space has not
been searched appropriately in these methods and thus they have low convergence
rate. For the first time, in this paper use a hyper-heuristic approach to find
an efficient feature subset. In the proposed method, genetic algorithm is used
as a supervisor and 16 heuristic algorithms are used as local searches.
Empirical study of the proposed method on several commonly used data sets from
UCI data sets indicates that it outperforms recent existing methods in the
literature for feature selection.
| [
{
"version": "v1",
"created": "Wed, 20 Jan 2016 20:59:55 GMT"
}
] | 2016-01-21T00:00:00 | [
[
"Montazeri",
"Mitra",
""
],
[
"Baghshah",
"Mahdieh Soleymani",
""
],
[
"Niknafs",
"Aliakbar",
""
]
] | TITLE: Selecting Efficient Features via a Hyper-Heuristic Approach
ABSTRACT: By Emerging huge databases and the need to efficient learning algorithms on
these datasets, new problems have appeared and some methods have been proposed
to solve these problems by selecting efficient features. Feature selection is a
problem of finding efficient features among all features in which the final
feature set can improve accuracy and reduce complexity. One way to solve this
problem is to evaluate all possible feature subsets. However, evaluating all
possible feature subsets is an exhaustive search and thus it has high
computational complexity. Until now many heuristic algorithms have been studied
for solving this problem. Hyper-heuristic is a new heuristic approach which can
search the solution space effectively by applying local searches appropriately.
Each local search is a neighborhood searching algorithm. Since each region of
the solution space can have its own characteristics, it should be chosen an
appropriate local search and apply it to current solution. This task is tackled
to a supervisor. The supervisor chooses a local search based on the functional
history of local searches. By doing this task, it can trade of between
exploitation and exploration. Since the existing heuristic cannot trade of
between exploration and exploitation appropriately, the solution space has not
been searched appropriately in these methods and thus they have low convergence
rate. For the first time, in this paper use a hyper-heuristic approach to find
an efficient feature subset. In the proposed method, genetic algorithm is used
as a supervisor and 16 heuristic algorithms are used as local searches.
Empirical study of the proposed method on several commonly used data sets from
UCI data sets indicates that it outperforms recent existing methods in the
literature for feature selection.
| no_new_dataset | 0.944022 |
1407.5245 | Liantao Wang | Ji Zhao, Liantao Wang, Ricardo Cabral, Fernando De la Torre | Feature and Region Selection for Visual Learning | null | IEEE Transactions on Image Processing, 2016, vol. 25, pp.
1084-1094 | 10.1109/TIP.2016.2514503 | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Visual learning problems such as object classification and action recognition
are typically approached using extensions of the popular bag-of-words (BoW)
model. Despite its great success, it is unclear what visual features the BoW
model is learning: Which regions in the image or video are used to discriminate
among classes? Which are the most discriminative visual words? Answering these
questions is fundamental for understanding existing BoW models and inspiring
better models for visual recognition.
To answer these questions, this paper presents a method for feature selection
and region selection in the visual BoW model. This allows for an intermediate
visualization of the features and regions that are important for visual
learning. The main idea is to assign latent weights to the features or regions,
and jointly optimize these latent variables with the parameters of a classifier
(e.g., support vector machine). There are four main benefits of our approach:
(1) Our approach accommodates non-linear additive kernels such as the popular
$\chi^2$ and intersection kernel; (2) our approach is able to handle both
regions in images and spatio-temporal regions in videos in a unified way; (3)
the feature selection problem is convex, and both problems can be solved using
a scalable reduced gradient method; (4) we point out strong connections with
multiple kernel learning and multiple instance learning approaches.
Experimental results in the PASCAL VOC 2007, MSR Action Dataset II and YouTube
illustrate the benefits of our approach.
| [
{
"version": "v1",
"created": "Sun, 20 Jul 2014 04:42:50 GMT"
},
{
"version": "v2",
"created": "Tue, 19 Jan 2016 03:27:59 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Zhao",
"Ji",
""
],
[
"Wang",
"Liantao",
""
],
[
"Cabral",
"Ricardo",
""
],
[
"De la Torre",
"Fernando",
""
]
] | TITLE: Feature and Region Selection for Visual Learning
ABSTRACT: Visual learning problems such as object classification and action recognition
are typically approached using extensions of the popular bag-of-words (BoW)
model. Despite its great success, it is unclear what visual features the BoW
model is learning: Which regions in the image or video are used to discriminate
among classes? Which are the most discriminative visual words? Answering these
questions is fundamental for understanding existing BoW models and inspiring
better models for visual recognition.
To answer these questions, this paper presents a method for feature selection
and region selection in the visual BoW model. This allows for an intermediate
visualization of the features and regions that are important for visual
learning. The main idea is to assign latent weights to the features or regions,
and jointly optimize these latent variables with the parameters of a classifier
(e.g., support vector machine). There are four main benefits of our approach:
(1) Our approach accommodates non-linear additive kernels such as the popular
$\chi^2$ and intersection kernel; (2) our approach is able to handle both
regions in images and spatio-temporal regions in videos in a unified way; (3)
the feature selection problem is convex, and both problems can be solved using
a scalable reduced gradient method; (4) we point out strong connections with
multiple kernel learning and multiple instance learning approaches.
Experimental results in the PASCAL VOC 2007, MSR Action Dataset II and YouTube
illustrate the benefits of our approach.
| no_new_dataset | 0.947575 |
1412.0826 | Chunhua Shen | Fumin Shen, Chunhua Shen, Qinfeng Shi, Anton van den Hengel, Zhenmin
Tang, Heng Tao Shen | Hashing on Nonlinear Manifolds | 13 pages. arXiv admin note: text overlap with arXiv:1303.7043 | null | 10.1109/TIP.2015.2405340 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning based hashing methods have attracted considerable attention due to
their ability to greatly increase the scale at which existing algorithms may
operate. Most of these methods are designed to generate binary codes preserving
the Euclidean similarity in the original space. Manifold learning techniques,
in contrast, are better able to model the intrinsic structure embedded in the
original high-dimensional data. The complexities of these models, and the
problems with out-of-sample data, have previously rendered them unsuitable for
application to large-scale embedding, however. In this work, how to learn
compact binary embeddings on their intrinsic manifolds is considered. In order
to address the above-mentioned difficulties, an efficient, inductive solution
to the out-of-sample data problem, and a process by which non-parametric
manifold learning may be used as the basis of a hashing method is proposed. The
proposed approach thus allows the development of a range of new hashing
techniques exploiting the flexibility of the wide variety of manifold learning
approaches available. It is particularly shown that hashing on the basis of
t-SNE outperforms state-of-the-art hashing methods on large-scale benchmark
datasets, and is very effective for image classification with very short code
lengths. The proposed hashing framework is shown to be easily improved, for
example, by minimizing the quantization error with learned orthogonal
rotations. In addition, a supervised inductive manifold hashing framework is
developed by incorporating the label information, which is shown to greatly
advance the semantic retrieval performance.
| [
{
"version": "v1",
"created": "Tue, 2 Dec 2014 09:36:12 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Shen",
"Fumin",
""
],
[
"Shen",
"Chunhua",
""
],
[
"Shi",
"Qinfeng",
""
],
[
"Hengel",
"Anton van den",
""
],
[
"Tang",
"Zhenmin",
""
],
[
"Shen",
"Heng Tao",
""
]
] | TITLE: Hashing on Nonlinear Manifolds
ABSTRACT: Learning based hashing methods have attracted considerable attention due to
their ability to greatly increase the scale at which existing algorithms may
operate. Most of these methods are designed to generate binary codes preserving
the Euclidean similarity in the original space. Manifold learning techniques,
in contrast, are better able to model the intrinsic structure embedded in the
original high-dimensional data. The complexities of these models, and the
problems with out-of-sample data, have previously rendered them unsuitable for
application to large-scale embedding, however. In this work, how to learn
compact binary embeddings on their intrinsic manifolds is considered. In order
to address the above-mentioned difficulties, an efficient, inductive solution
to the out-of-sample data problem, and a process by which non-parametric
manifold learning may be used as the basis of a hashing method is proposed. The
proposed approach thus allows the development of a range of new hashing
techniques exploiting the flexibility of the wide variety of manifold learning
approaches available. It is particularly shown that hashing on the basis of
t-SNE outperforms state-of-the-art hashing methods on large-scale benchmark
datasets, and is very effective for image classification with very short code
lengths. The proposed hashing framework is shown to be easily improved, for
example, by minimizing the quantization error with learned orthogonal
rotations. In addition, a supervised inductive manifold hashing framework is
developed by incorporating the label information, which is shown to greatly
advance the semantic retrieval performance.
| no_new_dataset | 0.947672 |
1505.00389 | Wangmeng Zuo | Zhaoxin Li, Kuanquan Wang, Wangmeng Zuo, Deyu Meng and Lei Zhang | Detail-preserving and Content-aware Variational Multi-view Stereo
Reconstruction | 14 pages,16 figures. Submitted to IEEE Transaction on image
processing | null | 10.1109/TIP.2015.2507400 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate recovery of 3D geometrical surfaces from calibrated 2D multi-view
images is a fundamental yet active research area in computer vision. Despite
the steady progress in multi-view stereo reconstruction, most existing methods
are still limited in recovering fine-scale details and sharp features while
suppressing noises, and may fail in reconstructing regions with few textures.
To address these limitations, this paper presents a Detail-preserving and
Content-aware Variational (DCV) multi-view stereo method, which reconstructs
the 3D surface by alternating between reprojection error minimization and mesh
denoising. In reprojection error minimization, we propose a novel inter-image
similarity measure, which is effective to preserve fine-scale details of the
reconstructed surface and builds a connection between guided image filtering
and image registration. In mesh denoising, we propose a content-aware
$\ell_{p}$-minimization algorithm by adaptively estimating the $p$ value and
regularization parameters based on the current input. It is much more promising
in suppressing noise while preserving sharp features than conventional
isotropic mesh smoothing. Experimental results on benchmark datasets
demonstrate that our DCV method is capable of recovering more surface details,
and obtains cleaner and more accurate reconstructions than state-of-the-art
methods. In particular, our method achieves the best results among all
published methods on the Middlebury dino ring and dino sparse ring datasets in
terms of both completeness and accuracy.
| [
{
"version": "v1",
"created": "Sun, 3 May 2015 03:03:49 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Li",
"Zhaoxin",
""
],
[
"Wang",
"Kuanquan",
""
],
[
"Zuo",
"Wangmeng",
""
],
[
"Meng",
"Deyu",
""
],
[
"Zhang",
"Lei",
""
]
] | TITLE: Detail-preserving and Content-aware Variational Multi-view Stereo
Reconstruction
ABSTRACT: Accurate recovery of 3D geometrical surfaces from calibrated 2D multi-view
images is a fundamental yet active research area in computer vision. Despite
the steady progress in multi-view stereo reconstruction, most existing methods
are still limited in recovering fine-scale details and sharp features while
suppressing noises, and may fail in reconstructing regions with few textures.
To address these limitations, this paper presents a Detail-preserving and
Content-aware Variational (DCV) multi-view stereo method, which reconstructs
the 3D surface by alternating between reprojection error minimization and mesh
denoising. In reprojection error minimization, we propose a novel inter-image
similarity measure, which is effective to preserve fine-scale details of the
reconstructed surface and builds a connection between guided image filtering
and image registration. In mesh denoising, we propose a content-aware
$\ell_{p}$-minimization algorithm by adaptively estimating the $p$ value and
regularization parameters based on the current input. It is much more promising
in suppressing noise while preserving sharp features than conventional
isotropic mesh smoothing. Experimental results on benchmark datasets
demonstrate that our DCV method is capable of recovering more surface details,
and obtains cleaner and more accurate reconstructions than state-of-the-art
methods. In particular, our method achieves the best results among all
published methods on the Middlebury dino ring and dino sparse ring datasets in
terms of both completeness and accuracy.
| no_new_dataset | 0.945901 |
1510.00132 | MIkhail Hushchyn | Mikhail Hushchyn, Philippe Charpentier, Andrey Ustyuzhanin | Disk storage management for LHCb based on Data Popularity estimator | null | null | 10.1088/1742-6596/664/4/042026 | null | cs.DC cs.LG physics.data-an | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents an algorithm providing recommendations for optimizing the
LHCb data storage. The LHCb data storage system is a hybrid system. All
datasets are kept as archives on magnetic tapes. The most popular datasets are
kept on disks. The algorithm takes the dataset usage history and metadata
(size, type, configuration etc.) to generate a recommendation report. This
article presents how we use machine learning algorithms to predict future data
popularity. Using these predictions it is possible to estimate which datasets
should be removed from disk. We use regression algorithms and time series
analysis to find the optimal number of replicas for datasets that are kept on
disk. Based on the data popularity and the number of replicas optimization, the
algorithm minimizes a loss function to find the optimal data distribution. The
loss function represents all requirements for data distribution in the data
storage system. We demonstrate how our algorithm helps to save disk space and
to reduce waiting times for jobs using this data.
| [
{
"version": "v1",
"created": "Thu, 1 Oct 2015 07:40:37 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Hushchyn",
"Mikhail",
""
],
[
"Charpentier",
"Philippe",
""
],
[
"Ustyuzhanin",
"Andrey",
""
]
] | TITLE: Disk storage management for LHCb based on Data Popularity estimator
ABSTRACT: This paper presents an algorithm providing recommendations for optimizing the
LHCb data storage. The LHCb data storage system is a hybrid system. All
datasets are kept as archives on magnetic tapes. The most popular datasets are
kept on disks. The algorithm takes the dataset usage history and metadata
(size, type, configuration etc.) to generate a recommendation report. This
article presents how we use machine learning algorithms to predict future data
popularity. Using these predictions it is possible to estimate which datasets
should be removed from disk. We use regression algorithms and time series
analysis to find the optimal number of replicas for datasets that are kept on
disk. Based on the data popularity and the number of replicas optimization, the
algorithm minimizes a loss function to find the optimal data distribution. The
loss function represents all requirements for data distribution in the data
storage system. We demonstrate how our algorithm helps to save disk space and
to reduce waiting times for jobs using this data.
| no_new_dataset | 0.951863 |
1510.00624 | Tatiana Likhomanenko | Tatiana Likhomanenko, Alex Rogozhnikov, Alexander Baranov, Egor
Khairullin, Andrey Ustyuzhanin | Reproducible Experiment Platform | 21st International Conference on Computing in High Energy Physics
(CHEP2015), 6 pages | null | 10.1088/1742-6596/664/5/052022 | null | physics.data-an | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data analysis in fundamental sciences nowadays is an essential process that
pushes frontiers of our knowledge and leads to new discoveries. At the same
time we can see that complexity of those analyses increases fast due to
a)~enormous volumes of datasets being analyzed, b)~variety of techniques and
algorithms one have to check inside a single analysis, c)~distributed nature of
research teams that requires special communication media for knowledge and
information exchange between individual researchers. There is a lot of
resemblance between techniques and problems arising in the areas of industrial
information retrieval and particle physics. To address those problems we
propose Reproducible Experiment Platform (REP), a software infrastructure to
support collaborative ecosystem for computational science. It is a Python based
solution for research teams that allows running computational experiments on
shared datasets, obtaining repeatable results, and consistent comparisons of
the obtained results. We present some key features of REP based on case studies
which include trigger optimization and physics analysis studies at the LHCb
experiment.
| [
{
"version": "v1",
"created": "Thu, 1 Oct 2015 11:41:08 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Likhomanenko",
"Tatiana",
""
],
[
"Rogozhnikov",
"Alex",
""
],
[
"Baranov",
"Alexander",
""
],
[
"Khairullin",
"Egor",
""
],
[
"Ustyuzhanin",
"Andrey",
""
]
] | TITLE: Reproducible Experiment Platform
ABSTRACT: Data analysis in fundamental sciences nowadays is an essential process that
pushes frontiers of our knowledge and leads to new discoveries. At the same
time we can see that complexity of those analyses increases fast due to
a)~enormous volumes of datasets being analyzed, b)~variety of techniques and
algorithms one have to check inside a single analysis, c)~distributed nature of
research teams that requires special communication media for knowledge and
information exchange between individual researchers. There is a lot of
resemblance between techniques and problems arising in the areas of industrial
information retrieval and particle physics. To address those problems we
propose Reproducible Experiment Platform (REP), a software infrastructure to
support collaborative ecosystem for computational science. It is a Python based
solution for research teams that allows running computational experiments on
shared datasets, obtaining repeatable results, and consistent comparisons of
the obtained results. We present some key features of REP based on case studies
which include trigger optimization and physics analysis studies at the LHCb
experiment.
| no_new_dataset | 0.941277 |
1510.07847 | Eugenio Valdano | Eugenio Valdano, Chiara Poletto, Vittoria Colizza | Infection propagator approach to compute epidemic thresholds on temporal
networks: impact of immunity and of limited temporal resolution | 23 pages, 8 figures | null | 10.1140/epjb/e2015-60620-5 | null | physics.soc-ph q-bio.PE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The epidemic threshold of a spreading process indicates the condition for the
occurrence of the wide spreading regime, thus representing a predictor of the
network vulnerability to the epidemic. Such threshold depends on the natural
history of the disease and on the pattern of contacts of the network with its
time variation. Based on the theoretical framework introduced in (Valdano et
al. PRX 2015) for a susceptible-infectious-susceptible model, we formulate here
an infection propagator approach to compute the epidemic threshold accounting
for more realistic effects regarding a varying force of infection per contact,
the presence of immunity, and a limited time resolution of the temporal
network. We apply the approach to two temporal network models and an empirical
dataset of school contacts. We find that permanent or temporary immunity do not
affect the estimation of the epidemic threshold through the infection
propagator approach. Comparisons with numerical results show the good agreement
of the analytical predictions. Aggregating the temporal network rapidly
deteriorates the predictions, except for slow diseases once the heterogeneity
of the links is preserved. Weight-topology correlations are found to be the
critical factor to be preserved to improve accuracy in the prediction.
| [
{
"version": "v1",
"created": "Tue, 27 Oct 2015 10:38:03 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Valdano",
"Eugenio",
""
],
[
"Poletto",
"Chiara",
""
],
[
"Colizza",
"Vittoria",
""
]
] | TITLE: Infection propagator approach to compute epidemic thresholds on temporal
networks: impact of immunity and of limited temporal resolution
ABSTRACT: The epidemic threshold of a spreading process indicates the condition for the
occurrence of the wide spreading regime, thus representing a predictor of the
network vulnerability to the epidemic. Such threshold depends on the natural
history of the disease and on the pattern of contacts of the network with its
time variation. Based on the theoretical framework introduced in (Valdano et
al. PRX 2015) for a susceptible-infectious-susceptible model, we formulate here
an infection propagator approach to compute the epidemic threshold accounting
for more realistic effects regarding a varying force of infection per contact,
the presence of immunity, and a limited time resolution of the temporal
network. We apply the approach to two temporal network models and an empirical
dataset of school contacts. We find that permanent or temporary immunity do not
affect the estimation of the epidemic threshold through the infection
propagator approach. Comparisons with numerical results show the good agreement
of the analytical predictions. Aggregating the temporal network rapidly
deteriorates the predictions, except for slow diseases once the heterogeneity
of the links is preserved. Weight-topology correlations are found to be the
critical factor to be preserved to improve accuracy in the prediction.
| no_new_dataset | 0.945851 |
1511.02919 | Deepti Ghadiyaram | Deepti Ghadiyaram and Alan C. Bovik | Massive Online Crowdsourced Study of Subjective and Objective Picture
Quality | 16 pages | null | 10.1109/TIP.2015.2500021 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most publicly available image quality databases have been created under
highly controlled conditions by introducing graded simulated distortions onto
high-quality photographs. However, images captured using typical real-world
mobile camera devices are usually afflicted by complex mixtures of multiple
distortions, which are not necessarily well-modeled by the synthetic
distortions found in existing databases. The originators of existing legacy
databases usually conducted human psychometric studies to obtain statistically
meaningful sets of human opinion scores on images in a stringently controlled
visual environment, resulting in small data collections relative to other kinds
of image analysis databases. Towards overcoming these limitations, we designed
and created a new database that we call the LIVE In the Wild Image Quality
Challenge Database, which contains widely diverse authentic image distortions
on a large number of images captured using a representative variety of modern
mobile devices. We also designed and implemented a new online crowdsourcing
system, which we have used to conduct a very large-scale, multi-month image
quality assessment subjective study. Our database consists of over 350000
opinion scores on 1162 images evaluated by over 7000 unique human observers.
Despite the lack of control over the experimental environments of the numerous
study participants, we demonstrate excellent internal consistency of the
subjective dataset. We also evaluate several top-performing blind Image Quality
Assessment algorithms on it and present insights on how mixtures of distortions
challenge both end users as well as automatic perceptual quality prediction
models.
| [
{
"version": "v1",
"created": "Mon, 9 Nov 2015 22:39:58 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Ghadiyaram",
"Deepti",
""
],
[
"Bovik",
"Alan C.",
""
]
] | TITLE: Massive Online Crowdsourced Study of Subjective and Objective Picture
Quality
ABSTRACT: Most publicly available image quality databases have been created under
highly controlled conditions by introducing graded simulated distortions onto
high-quality photographs. However, images captured using typical real-world
mobile camera devices are usually afflicted by complex mixtures of multiple
distortions, which are not necessarily well-modeled by the synthetic
distortions found in existing databases. The originators of existing legacy
databases usually conducted human psychometric studies to obtain statistically
meaningful sets of human opinion scores on images in a stringently controlled
visual environment, resulting in small data collections relative to other kinds
of image analysis databases. Towards overcoming these limitations, we designed
and created a new database that we call the LIVE In the Wild Image Quality
Challenge Database, which contains widely diverse authentic image distortions
on a large number of images captured using a representative variety of modern
mobile devices. We also designed and implemented a new online crowdsourcing
system, which we have used to conduct a very large-scale, multi-month image
quality assessment subjective study. Our database consists of over 350000
opinion scores on 1162 images evaluated by over 7000 unique human observers.
Despite the lack of control over the experimental environments of the numerous
study participants, we demonstrate excellent internal consistency of the
subjective dataset. We also evaluate several top-performing blind Image Quality
Assessment algorithms on it and present insights on how mixtures of distortions
challenge both end users as well as automatic perceptual quality prediction
models.
| new_dataset | 0.938237 |
1601.04745 | Xiaoxue Zhao | Xiaoxue Zhao, Jun Wang | A Theoretical Analysis of Two-Stage Recommendation for Cold-Start
Collaborative Filtering | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we present a theoretical framework for tackling the cold-start
collaborative filtering problem, where unknown targets (items or users) keep
coming to the system, and there is a limited number of resources (users or
items) that can be allocated and related to them. The solution requires a
trade-off between exploitation and exploration as with the limited
recommendation opportunities, we need to, on one hand, allocate the most
relevant resources right away, but, on the other hand, it is also necessary to
allocate resources that are useful for learning the target's properties in
order to recommend more relevant ones in the future. In this paper, we study a
simple two-stage recommendation combining a sequential and a batch solution
together. We first model the problem with the partially observable Markov
decision process (POMDP) and provide an exact solution. Then, through an
in-depth analysis over the POMDP value iteration solution, we identify that an
exact solution can be abstracted as selecting resources that are not only
highly relevant to the target according to the initial-stage information, but
also highly correlated, either positively or negatively, with other potential
resources for the next stage. With this finding, we propose an approximate
solution to ease the intractability of the exact solution. Our initial results
on synthetic data and the Movie Lens 100K dataset confirm the performance gains
of our theoretical development and analysis.
| [
{
"version": "v1",
"created": "Mon, 18 Jan 2016 22:31:06 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Zhao",
"Xiaoxue",
""
],
[
"Wang",
"Jun",
""
]
] | TITLE: A Theoretical Analysis of Two-Stage Recommendation for Cold-Start
Collaborative Filtering
ABSTRACT: In this paper, we present a theoretical framework for tackling the cold-start
collaborative filtering problem, where unknown targets (items or users) keep
coming to the system, and there is a limited number of resources (users or
items) that can be allocated and related to them. The solution requires a
trade-off between exploitation and exploration as with the limited
recommendation opportunities, we need to, on one hand, allocate the most
relevant resources right away, but, on the other hand, it is also necessary to
allocate resources that are useful for learning the target's properties in
order to recommend more relevant ones in the future. In this paper, we study a
simple two-stage recommendation combining a sequential and a batch solution
together. We first model the problem with the partially observable Markov
decision process (POMDP) and provide an exact solution. Then, through an
in-depth analysis over the POMDP value iteration solution, we identify that an
exact solution can be abstracted as selecting resources that are not only
highly relevant to the target according to the initial-stage information, but
also highly correlated, either positively or negatively, with other potential
resources for the next stage. With this finding, we propose an approximate
solution to ease the intractability of the exact solution. Our initial results
on synthetic data and the Movie Lens 100K dataset confirm the performance gains
of our theoretical development and analysis.
| no_new_dataset | 0.944125 |
1601.04800 | Zhao Kang | Zhao Kang, Chong Peng, Qiang Cheng | Top-N Recommender System via Matrix Completion | AAAI 2016 | null | null | null | cs.IR cs.AI cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Top-N recommender systems have been investigated widely both in industry and
academia. However, the recommendation quality is far from satisfactory. In this
paper, we propose a simple yet promising algorithm. We fill the user-item
matrix based on a low-rank assumption and simultaneously keep the original
information. To do that, a nonconvex rank relaxation rather than the nuclear
norm is adopted to provide a better rank approximation and an efficient
optimization strategy is designed. A comprehensive set of experiments on real
datasets demonstrates that our method pushes the accuracy of Top-N
recommendation to a new level.
| [
{
"version": "v1",
"created": "Tue, 19 Jan 2016 04:48:42 GMT"
}
] | 2016-01-20T00:00:00 | [
[
"Kang",
"Zhao",
""
],
[
"Peng",
"Chong",
""
],
[
"Cheng",
"Qiang",
""
]
] | TITLE: Top-N Recommender System via Matrix Completion
ABSTRACT: Top-N recommender systems have been investigated widely both in industry and
academia. However, the recommendation quality is far from satisfactory. In this
paper, we propose a simple yet promising algorithm. We fill the user-item
matrix based on a low-rank assumption and simultaneously keep the original
information. To do that, a nonconvex rank relaxation rather than the nuclear
norm is adopted to provide a better rank approximation and an efficient
optimization strategy is designed. A comprehensive set of experiments on real
datasets demonstrates that our method pushes the accuracy of Top-N
recommendation to a new level.
| no_new_dataset | 0.946941 |
1505.05192 | Carl Doersch | Carl Doersch and Abhinav Gupta and Alexei A. Efros | Unsupervised Visual Representation Learning by Context Prediction | Oral paper at ICCV 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work explores the use of spatial context as a source of free and
plentiful supervisory signal for training a rich visual representation. Given
only a large, unlabeled image collection, we extract random pairs of patches
from each image and train a convolutional neural net to predict the position of
the second patch relative to the first. We argue that doing well on this task
requires the model to learn to recognize objects and their parts. We
demonstrate that the feature representation learned using this within-image
context indeed captures visual similarity across images. For example, this
representation allows us to perform unsupervised visual discovery of objects
like cats, people, and even birds from the Pascal VOC 2011 detection dataset.
Furthermore, we show that the learned ConvNet can be used in the R-CNN
framework and provides a significant boost over a randomly-initialized ConvNet,
resulting in state-of-the-art performance among algorithms which use only
Pascal-provided training set annotations.
| [
{
"version": "v1",
"created": "Tue, 19 May 2015 21:18:17 GMT"
},
{
"version": "v2",
"created": "Mon, 28 Sep 2015 17:48:40 GMT"
},
{
"version": "v3",
"created": "Sat, 16 Jan 2016 22:09:45 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Doersch",
"Carl",
""
],
[
"Gupta",
"Abhinav",
""
],
[
"Efros",
"Alexei A.",
""
]
] | TITLE: Unsupervised Visual Representation Learning by Context Prediction
ABSTRACT: This work explores the use of spatial context as a source of free and
plentiful supervisory signal for training a rich visual representation. Given
only a large, unlabeled image collection, we extract random pairs of patches
from each image and train a convolutional neural net to predict the position of
the second patch relative to the first. We argue that doing well on this task
requires the model to learn to recognize objects and their parts. We
demonstrate that the feature representation learned using this within-image
context indeed captures visual similarity across images. For example, this
representation allows us to perform unsupervised visual discovery of objects
like cats, people, and even birds from the Pascal VOC 2011 detection dataset.
Furthermore, we show that the learned ConvNet can be used in the R-CNN
framework and provides a significant boost over a randomly-initialized ConvNet,
resulting in state-of-the-art performance among algorithms which use only
Pascal-provided training set annotations.
| no_new_dataset | 0.952442 |
1506.02158 | Yarin Gal | Yarin Gal, Zoubin Ghahramani | Bayesian Convolutional Neural Networks with Bernoulli Approximate
Variational Inference | 12 pages, 3 figures, ICLR format, updated with reviewer comments | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Convolutional neural networks (CNNs) work well on large datasets. But
labelled data is hard to collect, and in some applications larger amounts of
data are not available. The problem then is how to use CNNs with small data --
as CNNs overfit quickly. We present an efficient Bayesian CNN, offering better
robustness to over-fitting on small data than traditional approaches. This is
by placing a probability distribution over the CNN's kernels. We approximate
our model's intractable posterior with Bernoulli variational distributions,
requiring no additional model parameters.
On the theoretical side, we cast dropout network training as approximate
inference in Bayesian neural networks. This allows us to implement our model
using existing tools in deep learning with no increase in time complexity,
while highlighting a negative result in the field. We show a considerable
improvement in classification accuracy compared to standard techniques and
improve on published state-of-the-art results for CIFAR-10.
| [
{
"version": "v1",
"created": "Sat, 6 Jun 2015 14:43:40 GMT"
},
{
"version": "v2",
"created": "Thu, 27 Aug 2015 13:30:17 GMT"
},
{
"version": "v3",
"created": "Sun, 27 Sep 2015 13:34:58 GMT"
},
{
"version": "v4",
"created": "Mon, 2 Nov 2015 14:33:59 GMT"
},
{
"version": "v5",
"created": "Mon, 30 Nov 2015 21:22:15 GMT"
},
{
"version": "v6",
"created": "Mon, 18 Jan 2016 20:42:07 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Gal",
"Yarin",
""
],
[
"Ghahramani",
"Zoubin",
""
]
] | TITLE: Bayesian Convolutional Neural Networks with Bernoulli Approximate
Variational Inference
ABSTRACT: Convolutional neural networks (CNNs) work well on large datasets. But
labelled data is hard to collect, and in some applications larger amounts of
data are not available. The problem then is how to use CNNs with small data --
as CNNs overfit quickly. We present an efficient Bayesian CNN, offering better
robustness to over-fitting on small data than traditional approaches. This is
by placing a probability distribution over the CNN's kernels. We approximate
our model's intractable posterior with Bernoulli variational distributions,
requiring no additional model parameters.
On the theoretical side, we cast dropout network training as approximate
inference in Bayesian neural networks. This allows us to implement our model
using existing tools in deep learning with no increase in time complexity,
while highlighting a negative result in the field. We show a considerable
improvement in classification accuracy compared to standard techniques and
improve on published state-of-the-art results for CIFAR-10.
| no_new_dataset | 0.949059 |
1511.02554 | Hojjat Salehinejad | Farhad Pouladi, Hojjat Salehinejad and Amir Mohammad Gilani | Deep Recurrent Neural Networks for Sequential Phenotype Prediction in
Genomics | The articles is accepted at DeSE 2015 | null | null | null | cs.NE cs.CE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In analyzing of modern biological data, we are often dealing with ill-posed
problems and missing data, mostly due to high dimensionality and
multicollinearity of the dataset. In this paper, we have proposed a system
based on matrix factorization (MF) and deep recurrent neural networks (DRNNs)
for genotype imputation and phenotype sequences prediction. In order to model
the long-term dependencies of phenotype data, the new Recurrent Linear Units
(ReLU) learning strategy is utilized for the first time. The proposed model is
implemented for parallel processing on central processing units (CPUs) and
graphic processing units (GPUs). Performance of the proposed model is compared
with other training algorithms for learning long-term dependencies as well as
the sparse partial least square (SPLS) method on a set of genotype and
phenotype data with 604 samples, 1980 single-nucleotide polymorphisms (SNPs),
and two traits. The results demonstrate performance of the ReLU training
algorithm in learning long-term dependencies in RNNs.
| [
{
"version": "v1",
"created": "Mon, 9 Nov 2015 02:11:00 GMT"
},
{
"version": "v2",
"created": "Tue, 1 Dec 2015 20:48:34 GMT"
},
{
"version": "v3",
"created": "Sun, 17 Jan 2016 03:30:10 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Pouladi",
"Farhad",
""
],
[
"Salehinejad",
"Hojjat",
""
],
[
"Gilani",
"Amir Mohammad",
""
]
] | TITLE: Deep Recurrent Neural Networks for Sequential Phenotype Prediction in
Genomics
ABSTRACT: In analyzing of modern biological data, we are often dealing with ill-posed
problems and missing data, mostly due to high dimensionality and
multicollinearity of the dataset. In this paper, we have proposed a system
based on matrix factorization (MF) and deep recurrent neural networks (DRNNs)
for genotype imputation and phenotype sequences prediction. In order to model
the long-term dependencies of phenotype data, the new Recurrent Linear Units
(ReLU) learning strategy is utilized for the first time. The proposed model is
implemented for parallel processing on central processing units (CPUs) and
graphic processing units (GPUs). Performance of the proposed model is compared
with other training algorithms for learning long-term dependencies as well as
the sparse partial least square (SPLS) method on a set of genotype and
phenotype data with 604 samples, 1980 single-nucleotide polymorphisms (SNPs),
and two traits. The results demonstrate performance of the ReLU training
algorithm in learning long-term dependencies in RNNs.
| no_new_dataset | 0.950595 |
1512.02752 | Qi Mao | Qi Mao, Li Wang, Ivor W. Tsang, Yijun Sun | A Novel Regularized Principal Graph Learning Framework on Explicit Graph
Representation | null | null | null | null | cs.AI cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many scientific datasets are of high dimension, and the analysis usually
requires visual manipulation by retaining the most important structures of
data. Principal curve is a widely used approach for this purpose. However, many
existing methods work only for data with structures that are not
self-intersected, which is quite restrictive for real applications. A few
methods can overcome the above problem, but they either require complicated
human-made rules for a specific task with lack of convergence guarantee and
adaption flexibility to different tasks, or cannot obtain explicit structures
of data. To address these issues, we develop a new regularized principal graph
learning framework that captures the local information of the underlying graph
structure based on reversed graph embedding. As showcases, models that can
learn a spanning tree or a weighted undirected $\ell_1$ graph are proposed, and
a new learning algorithm is developed that learns a set of principal points and
a graph structure from data, simultaneously. The new algorithm is simple with
guaranteed convergence. We then extend the proposed framework to deal with
large-scale data. Experimental results on various synthetic and six real world
datasets show that the proposed method compares favorably with baselines and
can uncover the underlying structure correctly.
| [
{
"version": "v1",
"created": "Wed, 9 Dec 2015 04:57:18 GMT"
},
{
"version": "v2",
"created": "Sun, 17 Jan 2016 14:34:14 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Mao",
"Qi",
""
],
[
"Wang",
"Li",
""
],
[
"Tsang",
"Ivor W.",
""
],
[
"Sun",
"Yijun",
""
]
] | TITLE: A Novel Regularized Principal Graph Learning Framework on Explicit Graph
Representation
ABSTRACT: Many scientific datasets are of high dimension, and the analysis usually
requires visual manipulation by retaining the most important structures of
data. Principal curve is a widely used approach for this purpose. However, many
existing methods work only for data with structures that are not
self-intersected, which is quite restrictive for real applications. A few
methods can overcome the above problem, but they either require complicated
human-made rules for a specific task with lack of convergence guarantee and
adaption flexibility to different tasks, or cannot obtain explicit structures
of data. To address these issues, we develop a new regularized principal graph
learning framework that captures the local information of the underlying graph
structure based on reversed graph embedding. As showcases, models that can
learn a spanning tree or a weighted undirected $\ell_1$ graph are proposed, and
a new learning algorithm is developed that learns a set of principal points and
a graph structure from data, simultaneously. The new algorithm is simple with
guaranteed convergence. We then extend the proposed framework to deal with
large-scale data. Experimental results on various synthetic and six real world
datasets show that the proposed method compares favorably with baselines and
can uncover the underlying structure correctly.
| no_new_dataset | 0.944331 |
1512.08120 | Fanhua Shang | Fanhua Shang and James Cheng and Hong Cheng | Regularized Orthogonal Tensor Decompositions for Multi-Relational
Learning | 18 pages, 10 figures | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multi-relational learning has received lots of attention from researchers in
various research communities. Most existing methods either suffer from
superlinear per-iteration cost, or are sensitive to the given ranks. To address
both issues, we propose a scalable core tensor trace norm Regularized
Orthogonal Iteration Decomposition (ROID) method for full or incomplete tensor
analytics, which can be generalized as a graph Laplacian regularized version by
using auxiliary information or a sparse higher-order orthogonal iteration
(SHOOI) version. We first induce the equivalence relation of the Schatten
p-norm (0<p<\infty) of a low multi-linear rank tensor and its core tensor. Then
we achieve a much smaller matrix trace norm minimization problem. Finally, we
develop two efficient augmented Lagrange multiplier algorithms to solve our
problems with convergence guarantees. Extensive experiments using both real and
synthetic datasets, even though with only a few observations, verified both the
efficiency and effectiveness of our methods.
| [
{
"version": "v1",
"created": "Sat, 26 Dec 2015 15:26:05 GMT"
},
{
"version": "v2",
"created": "Sat, 16 Jan 2016 15:32:15 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Shang",
"Fanhua",
""
],
[
"Cheng",
"James",
""
],
[
"Cheng",
"Hong",
""
]
] | TITLE: Regularized Orthogonal Tensor Decompositions for Multi-Relational
Learning
ABSTRACT: Multi-relational learning has received lots of attention from researchers in
various research communities. Most existing methods either suffer from
superlinear per-iteration cost, or are sensitive to the given ranks. To address
both issues, we propose a scalable core tensor trace norm Regularized
Orthogonal Iteration Decomposition (ROID) method for full or incomplete tensor
analytics, which can be generalized as a graph Laplacian regularized version by
using auxiliary information or a sparse higher-order orthogonal iteration
(SHOOI) version. We first induce the equivalence relation of the Schatten
p-norm (0<p<\infty) of a low multi-linear rank tensor and its core tensor. Then
we achieve a much smaller matrix trace norm minimization problem. Finally, we
develop two efficient augmented Lagrange multiplier algorithms to solve our
problems with convergence guarantees. Extensive experiments using both real and
synthetic datasets, even though with only a few observations, verified both the
efficiency and effectiveness of our methods.
| no_new_dataset | 0.947186 |
1601.04386 | Ying Huang | Ying Huang, Hong Zheng, Haibin Ling, Erik Blasch, Hao Yang | A Comparative Study of Object Trackers for Infrared Flying Bird Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Bird strikes present a huge risk for aircraft, especially since traditional
airport bird surveillance is mainly dependent on inefficient human observation.
Computer vision based technology has been proposed to automatically detect
birds, determine bird flying trajectories, and predict aircraft takeoff delays.
However, the characteristics of bird flight using imagery and the performance
of existing methods applied to flying bird task are not well known. Therefore,
we perform infrared flying bird tracking experiments using 12 state-of-the-art
algorithms on a real BIRDSITE-IR dataset to obtain useful clues and recommend
feature analysis. We also develop a Struck-scale method to demonstrate the
effectiveness of multiple scale sampling adaption in handling the object of
flying bird with varying shape and scale. The general analysis can be used to
develop specialized bird tracking methods for airport safety, wildness and
urban bird population studies.
| [
{
"version": "v1",
"created": "Mon, 18 Jan 2016 02:08:18 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Huang",
"Ying",
""
],
[
"Zheng",
"Hong",
""
],
[
"Ling",
"Haibin",
""
],
[
"Blasch",
"Erik",
""
],
[
"Yang",
"Hao",
""
]
] | TITLE: A Comparative Study of Object Trackers for Infrared Flying Bird Tracking
ABSTRACT: Bird strikes present a huge risk for aircraft, especially since traditional
airport bird surveillance is mainly dependent on inefficient human observation.
Computer vision based technology has been proposed to automatically detect
birds, determine bird flying trajectories, and predict aircraft takeoff delays.
However, the characteristics of bird flight using imagery and the performance
of existing methods applied to flying bird task are not well known. Therefore,
we perform infrared flying bird tracking experiments using 12 state-of-the-art
algorithms on a real BIRDSITE-IR dataset to obtain useful clues and recommend
feature analysis. We also develop a Struck-scale method to demonstrate the
effectiveness of multiple scale sampling adaption in handling the object of
flying bird with varying shape and scale. The general analysis can be used to
develop specialized bird tracking methods for airport safety, wildness and
urban bird population studies.
| no_new_dataset | 0.928603 |
1601.04406 | Vinay Bettadapura | Vinay Bettadapura, Daniel Castro, Irfan Essa | Discovering Picturesque Highlights from Egocentric Vacation Videos | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present an approach for identifying picturesque highlights from large
amounts of egocentric video data. Given a set of egocentric videos captured
over the course of a vacation, our method analyzes the videos and looks for
images that have good picturesque and artistic properties. We introduce novel
techniques to automatically determine aesthetic features such as composition,
symmetry and color vibrancy in egocentric videos and rank the video frames
based on their photographic qualities to generate highlights. Our approach also
uses contextual information such as GPS, when available, to assess the relative
importance of each geographic location where the vacation videos were shot.
Furthermore, we specifically leverage the properties of egocentric videos to
improve our highlight detection. We demonstrate results on a new egocentric
vacation dataset which includes 26.5 hours of videos taken over a 14 day
vacation that spans many famous tourist destinations and also provide results
from a user-study to access our results.
| [
{
"version": "v1",
"created": "Mon, 18 Jan 2016 06:23:14 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Bettadapura",
"Vinay",
""
],
[
"Castro",
"Daniel",
""
],
[
"Essa",
"Irfan",
""
]
] | TITLE: Discovering Picturesque Highlights from Egocentric Vacation Videos
ABSTRACT: We present an approach for identifying picturesque highlights from large
amounts of egocentric video data. Given a set of egocentric videos captured
over the course of a vacation, our method analyzes the videos and looks for
images that have good picturesque and artistic properties. We introduce novel
techniques to automatically determine aesthetic features such as composition,
symmetry and color vibrancy in egocentric videos and rank the video frames
based on their photographic qualities to generate highlights. Our approach also
uses contextual information such as GPS, when available, to assess the relative
importance of each geographic location where the vacation videos were shot.
Furthermore, we specifically leverage the properties of egocentric videos to
improve our highlight detection. We demonstrate results on a new egocentric
vacation dataset which includes 26.5 hours of videos taken over a 14 day
vacation that spans many famous tourist destinations and also provide results
from a user-study to access our results.
| new_dataset | 0.953275 |
1601.04602 | Kevin Taylor-Sakyi | Kevin Taylor-Sakyi | Big Data: Understanding Big Data | 8 pages, Big Data Analytics, Data Storage, MapReduce,
Knowledge-Space, Big Data Inconsistencies | null | null | null | cs.DC cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Steve Jobs, one of the greatest visionaries of our time was quoted in 1996
saying "a lot of times, people do not know what they want until you show it to
them" [38] indicating he advocated products to be developed based on human
intuition rather than research. With the advancements of mobile devices, social
networks and the Internet of Things, enormous amounts of complex data, both
structured and unstructured are being captured in hope to allow organizations
to make better business decisions as data is now vital for an organizations
success. These enormous amounts of data are referred to as Big Data, which
enables a competitive advantage over rivals when processed and analyzed
appropriately. However Big Data Analytics has a few concerns including
Management of Data-lifecycle, Privacy & Security, and Data Representation. This
paper reviews the fundamental concept of Big Data, the Data Storage domain, the
MapReduce programming paradigm used in processing these large datasets, and
focuses on two case studies showing the effectiveness of Big Data Analytics and
presents how it could be of greater good in the future if handled
appropriately.
| [
{
"version": "v1",
"created": "Fri, 15 Jan 2016 19:10:43 GMT"
}
] | 2016-01-19T00:00:00 | [
[
"Taylor-Sakyi",
"Kevin",
""
]
] | TITLE: Big Data: Understanding Big Data
ABSTRACT: Steve Jobs, one of the greatest visionaries of our time was quoted in 1996
saying "a lot of times, people do not know what they want until you show it to
them" [38] indicating he advocated products to be developed based on human
intuition rather than research. With the advancements of mobile devices, social
networks and the Internet of Things, enormous amounts of complex data, both
structured and unstructured are being captured in hope to allow organizations
to make better business decisions as data is now vital for an organizations
success. These enormous amounts of data are referred to as Big Data, which
enables a competitive advantage over rivals when processed and analyzed
appropriately. However Big Data Analytics has a few concerns including
Management of Data-lifecycle, Privacy & Security, and Data Representation. This
paper reviews the fundamental concept of Big Data, the Data Storage domain, the
MapReduce programming paradigm used in processing these large datasets, and
focuses on two case studies showing the effectiveness of Big Data Analytics and
presents how it could be of greater good in the future if handled
appropriately.
| no_new_dataset | 0.948155 |
1502.04434 | Sergey Demyanov | Sergey Demyanov, James Bailey, Ramamohanarao Kotagiri, Christopher
Leckie | Invariant backpropagation: how to train a transformation-invariant
neural network | null | null | null | null | stat.ML cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In many classification problems a classifier should be robust to small
variations in the input vector. This is a desired property not only for
particular transformations, such as translation and rotation in image
classification problems, but also for all others for which the change is small
enough to retain the object perceptually indistinguishable. We propose two
extensions of the backpropagation algorithm that train a neural network to be
robust to variations in the feature vector. While the first of them enforces
robustness of the loss function to all variations, the second method trains the
predictions to be robust to a particular variation which changes the loss
function the most. The second methods demonstrates better results, but is
slightly slower. We analytically compare the proposed algorithm with two the
most similar approaches (Tangent BP and Adversarial Training), and propose
their fast versions. In the experimental part we perform comparison of all
algorithms in terms of classification accuracy and robustness to noise on MNIST
and CIFAR-10 datasets. Additionally we analyze how the performance of the
proposed algorithm depends on the dataset size and data augmentation.
| [
{
"version": "v1",
"created": "Mon, 16 Feb 2015 06:28:35 GMT"
},
{
"version": "v2",
"created": "Mon, 2 Nov 2015 11:44:59 GMT"
},
{
"version": "v3",
"created": "Fri, 15 Jan 2016 04:49:00 GMT"
}
] | 2016-01-18T00:00:00 | [
[
"Demyanov",
"Sergey",
""
],
[
"Bailey",
"James",
""
],
[
"Kotagiri",
"Ramamohanarao",
""
],
[
"Leckie",
"Christopher",
""
]
] | TITLE: Invariant backpropagation: how to train a transformation-invariant
neural network
ABSTRACT: In many classification problems a classifier should be robust to small
variations in the input vector. This is a desired property not only for
particular transformations, such as translation and rotation in image
classification problems, but also for all others for which the change is small
enough to retain the object perceptually indistinguishable. We propose two
extensions of the backpropagation algorithm that train a neural network to be
robust to variations in the feature vector. While the first of them enforces
robustness of the loss function to all variations, the second method trains the
predictions to be robust to a particular variation which changes the loss
function the most. The second methods demonstrates better results, but is
slightly slower. We analytically compare the proposed algorithm with two the
most similar approaches (Tangent BP and Adversarial Training), and propose
their fast versions. In the experimental part we perform comparison of all
algorithms in terms of classification accuracy and robustness to noise on MNIST
and CIFAR-10 datasets. Additionally we analyze how the performance of the
proposed algorithm depends on the dataset size and data augmentation.
| no_new_dataset | 0.948202 |
1601.03229 | Jun Zhang | Jun Zhang and Xiaokui Xiao and Xing Xie | PrivTree: A Differentially Private Algorithm for Hierarchical
Decompositions | A short version of this paper will appear in SIGMOD 2016 | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given a set D of tuples defined on a domain Omega, we study differentially
private algorithms for constructing a histogram over Omega to approximate the
tuple distribution in D. Existing solutions for the problem mostly adopt a
hierarchical decomposition approach, which recursively splits Omega into
sub-domains and computes a noisy tuple count for each sub-domain, until all
noisy counts are below a certain threshold. This approach, however, requires
that we (i) impose a limit h on the recursion depth in the splitting of Omega
and (ii) set the noise in each count to be proportional to h. This leads to
inferior data utility due to the following dilemma: if we use a small h, then
the resulting histogram would be too coarse-grained to provide an accurate
approximation of data distribution; meanwhile, a large h would yield a
fine-grained histogram, but its quality would be severely degraded by the
increased amount of noise in the tuple counts.
To remedy the deficiency of existing solutions, we present PrivTree, a
histogram construction algorithm that also applies hierarchical decomposition
but features a crucial (and somewhat surprising) improvement: when deciding
whether or not to split a sub-domain, the amount of noise required in the
corresponding tuple count is independent of the recursive depth. This enables
PrivTree to adaptively generate high-quality histograms without even asking for
a pre-defined threshold on the depth of sub-domain splitting. As concrete
examples, we demonstrate an application of PrivTree in modelling spatial data,
and show that it can also be extended to handle sequence data (where the
decision in sub-domain splitting is not based on tuple counts but a more
sophisticated measure). Our experiments on a variety of real datasets show that
PrivTree significantly outperforms the states of the art in terms of data
utility.
| [
{
"version": "v1",
"created": "Wed, 13 Jan 2016 13:17:08 GMT"
},
{
"version": "v2",
"created": "Fri, 15 Jan 2016 02:51:31 GMT"
}
] | 2016-01-18T00:00:00 | [
[
"Zhang",
"Jun",
""
],
[
"Xiao",
"Xiaokui",
""
],
[
"Xie",
"Xing",
""
]
] | TITLE: PrivTree: A Differentially Private Algorithm for Hierarchical
Decompositions
ABSTRACT: Given a set D of tuples defined on a domain Omega, we study differentially
private algorithms for constructing a histogram over Omega to approximate the
tuple distribution in D. Existing solutions for the problem mostly adopt a
hierarchical decomposition approach, which recursively splits Omega into
sub-domains and computes a noisy tuple count for each sub-domain, until all
noisy counts are below a certain threshold. This approach, however, requires
that we (i) impose a limit h on the recursion depth in the splitting of Omega
and (ii) set the noise in each count to be proportional to h. This leads to
inferior data utility due to the following dilemma: if we use a small h, then
the resulting histogram would be too coarse-grained to provide an accurate
approximation of data distribution; meanwhile, a large h would yield a
fine-grained histogram, but its quality would be severely degraded by the
increased amount of noise in the tuple counts.
To remedy the deficiency of existing solutions, we present PrivTree, a
histogram construction algorithm that also applies hierarchical decomposition
but features a crucial (and somewhat surprising) improvement: when deciding
whether or not to split a sub-domain, the amount of noise required in the
corresponding tuple count is independent of the recursive depth. This enables
PrivTree to adaptively generate high-quality histograms without even asking for
a pre-defined threshold on the depth of sub-domain splitting. As concrete
examples, we demonstrate an application of PrivTree in modelling spatial data,
and show that it can also be extended to handle sequence data (where the
decision in sub-domain splitting is not based on tuple counts but a more
sophisticated measure). Our experiments on a variety of real datasets show that
PrivTree significantly outperforms the states of the art in terms of data
utility.
| no_new_dataset | 0.949248 |
1601.03754 | Ryan Curtin | Ryan R. Curtin | Dual-tree $k$-means with bounded iteration runtime | supplementary material included; submitted to ICML '16 | null | null | null | cs.DS cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | k-means is a widely used clustering algorithm, but for $k$ clusters and a
dataset size of $N$, each iteration of Lloyd's algorithm costs $O(kN)$ time.
Although there are existing techniques to accelerate single Lloyd iterations,
none of these are tailored to the case of large $k$, which is increasingly
common as dataset sizes grow. We propose a dual-tree algorithm that gives the
exact same results as standard $k$-means; when using cover trees, we use
adaptive analysis techniques to, under some assumptions, bound the
single-iteration runtime of the algorithm as $O(N + k log k)$. To our knowledge
these are the first sub-$O(kN)$ bounds for exact Lloyd iterations. We then show
that this theoretically favorable algorithm performs competitively in practice,
especially for large $N$ and $k$ in low dimensions. Further, the algorithm is
tree-independent, so any type of tree may be used.
| [
{
"version": "v1",
"created": "Thu, 14 Jan 2016 21:18:06 GMT"
}
] | 2016-01-18T00:00:00 | [
[
"Curtin",
"Ryan R.",
""
]
] | TITLE: Dual-tree $k$-means with bounded iteration runtime
ABSTRACT: k-means is a widely used clustering algorithm, but for $k$ clusters and a
dataset size of $N$, each iteration of Lloyd's algorithm costs $O(kN)$ time.
Although there are existing techniques to accelerate single Lloyd iterations,
none of these are tailored to the case of large $k$, which is increasingly
common as dataset sizes grow. We propose a dual-tree algorithm that gives the
exact same results as standard $k$-means; when using cover trees, we use
adaptive analysis techniques to, under some assumptions, bound the
single-iteration runtime of the algorithm as $O(N + k log k)$. To our knowledge
these are the first sub-$O(kN)$ bounds for exact Lloyd iterations. We then show
that this theoretically favorable algorithm performs competitively in practice,
especially for large $N$ and $k$ in low dimensions. Further, the algorithm is
tree-independent, so any type of tree may be used.
| no_new_dataset | 0.945951 |
1601.03797 | Sanjay Krishnan | Sanjay Krishnan, Jiannan Wang, Eugene Wu, Michael J. Franklin, Ken
Goldberg | ActiveClean: Interactive Data Cleaning While Learning Convex Loss Models | Pre-print | null | null | null | cs.DB cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data cleaning is often an important step to ensure that predictive models,
such as regression and classification, are not affected by systematic errors
such as inconsistent, out-of-date, or outlier data. Identifying dirty data is
often a manual and iterative process, and can be challenging on large datasets.
However, many data cleaning workflows can introduce subtle biases into the
training processes due to violation of independence assumptions. We propose
ActiveClean, a progressive cleaning approach where the model is updated
incrementally instead of re-training and can guarantee accuracy on partially
cleaned data. ActiveClean supports a popular class of models called convex loss
models (e.g., linear regression and SVMs). ActiveClean also leverages the
structure of a user's model to prioritize cleaning those records likely to
affect the results. We evaluate ActiveClean on five real-world datasets UCI
Adult, UCI EEG, MNIST, Dollars For Docs, and WorldBank with both real and
synthetic errors. Our results suggest that our proposed optimizations can
improve model accuracy by up-to 2.5x for the same amount of data cleaned.
Furthermore for a fixed cleaning budget and on all real dirty datasets,
ActiveClean returns more accurate models than uniform sampling and Active
Learning.
| [
{
"version": "v1",
"created": "Fri, 15 Jan 2016 02:02:00 GMT"
}
] | 2016-01-18T00:00:00 | [
[
"Krishnan",
"Sanjay",
""
],
[
"Wang",
"Jiannan",
""
],
[
"Wu",
"Eugene",
""
],
[
"Franklin",
"Michael J.",
""
],
[
"Goldberg",
"Ken",
""
]
] | TITLE: ActiveClean: Interactive Data Cleaning While Learning Convex Loss Models
ABSTRACT: Data cleaning is often an important step to ensure that predictive models,
such as regression and classification, are not affected by systematic errors
such as inconsistent, out-of-date, or outlier data. Identifying dirty data is
often a manual and iterative process, and can be challenging on large datasets.
However, many data cleaning workflows can introduce subtle biases into the
training processes due to violation of independence assumptions. We propose
ActiveClean, a progressive cleaning approach where the model is updated
incrementally instead of re-training and can guarantee accuracy on partially
cleaned data. ActiveClean supports a popular class of models called convex loss
models (e.g., linear regression and SVMs). ActiveClean also leverages the
structure of a user's model to prioritize cleaning those records likely to
affect the results. We evaluate ActiveClean on five real-world datasets UCI
Adult, UCI EEG, MNIST, Dollars For Docs, and WorldBank with both real and
synthetic errors. Our results suggest that our proposed optimizations can
improve model accuracy by up-to 2.5x for the same amount of data cleaned.
Furthermore for a fixed cleaning budget and on all real dirty datasets,
ActiveClean returns more accurate models than uniform sampling and Active
Learning.
| no_new_dataset | 0.945601 |
1504.03293 | Yuting Zhang | Yuting Zhang, Kihyuk Sohn, Ruben Villegas, Gang Pan, Honglak Lee | Improving Object Detection with Deep Convolutional Networks via Bayesian
Optimization and Structured Prediction | CVPR 2015 | null | 10.1109/CVPR.2015.7298621 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Object detection systems based on the deep convolutional neural network (CNN)
have recently made ground- breaking advances on several object detection
benchmarks. While the features learned by these high-capacity neural networks
are discriminative for categorization, inaccurate localization is still a major
source of error for detection. Building upon high-capacity CNN architectures,
we address the localization problem by 1) using a search algorithm based on
Bayesian optimization that sequentially proposes candidate regions for an
object bounding box, and 2) training the CNN with a structured loss that
explicitly penalizes the localization inaccuracy. In experiments, we
demonstrated that each of the proposed methods improves the detection
performance over the baseline method on PASCAL VOC 2007 and 2012 datasets.
Furthermore, two methods are complementary and significantly outperform the
previous state-of-the-art when combined.
| [
{
"version": "v1",
"created": "Mon, 13 Apr 2015 18:50:51 GMT"
},
{
"version": "v2",
"created": "Wed, 13 Jan 2016 18:27:32 GMT"
},
{
"version": "v3",
"created": "Thu, 14 Jan 2016 04:11:45 GMT"
}
] | 2016-01-15T00:00:00 | [
[
"Zhang",
"Yuting",
""
],
[
"Sohn",
"Kihyuk",
""
],
[
"Villegas",
"Ruben",
""
],
[
"Pan",
"Gang",
""
],
[
"Lee",
"Honglak",
""
]
] | TITLE: Improving Object Detection with Deep Convolutional Networks via Bayesian
Optimization and Structured Prediction
ABSTRACT: Object detection systems based on the deep convolutional neural network (CNN)
have recently made ground- breaking advances on several object detection
benchmarks. While the features learned by these high-capacity neural networks
are discriminative for categorization, inaccurate localization is still a major
source of error for detection. Building upon high-capacity CNN architectures,
we address the localization problem by 1) using a search algorithm based on
Bayesian optimization that sequentially proposes candidate regions for an
object bounding box, and 2) training the CNN with a structured loss that
explicitly penalizes the localization inaccuracy. In experiments, we
demonstrated that each of the proposed methods improves the detection
performance over the baseline method on PASCAL VOC 2007 and 2012 datasets.
Furthermore, two methods are complementary and significantly outperform the
previous state-of-the-art when combined.
| no_new_dataset | 0.950273 |
1510.03710 | Miroslav Vodol\'an | Miroslav Vodol\'an and Rudolf Kadlec and Jan Kleindienst | Hybrid Dialog State Tracker | Accepted to Machine Learning for SLU & Interaction NIPS 2015
Workshop. Model description in Section 2.1 simplified compared to the
previous version | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a hybrid dialog state tracker that combines a rule based
and a machine learning based approach to belief state tracking. Therefore, we
call it a hybrid tracker. The machine learning in our tracker is realized by a
Long Short Term Memory (LSTM) network. To our knowledge, our hybrid tracker
sets a new state-of-the-art result for the Dialog State Tracking Challenge
(DSTC) 2 dataset when the system uses only live SLU as its input.
| [
{
"version": "v1",
"created": "Tue, 13 Oct 2015 14:44:01 GMT"
},
{
"version": "v2",
"created": "Tue, 3 Nov 2015 08:38:14 GMT"
},
{
"version": "v3",
"created": "Thu, 14 Jan 2016 10:40:31 GMT"
}
] | 2016-01-15T00:00:00 | [
[
"Vodolán",
"Miroslav",
""
],
[
"Kadlec",
"Rudolf",
""
],
[
"Kleindienst",
"Jan",
""
]
] | TITLE: Hybrid Dialog State Tracker
ABSTRACT: This paper presents a hybrid dialog state tracker that combines a rule based
and a machine learning based approach to belief state tracking. Therefore, we
call it a hybrid tracker. The machine learning in our tracker is realized by a
Long Short Term Memory (LSTM) network. To our knowledge, our hybrid tracker
sets a new state-of-the-art result for the Dialog State Tracking Challenge
(DSTC) 2 dataset when the system uses only live SLU as its input.
| no_new_dataset | 0.940463 |
1511.00898 | Jouni Sir\'en | Jouni Sir\'en | Burrows-Wheeler transform for terabases | This is the full version of the paper that was accepted to DCC 2016.
The implementation is available at https://github.com/jltsiren/bwt-merge | null | null | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In order to avoid the reference bias introduced by mapping reads to a
reference genome, bioinformaticians are investigating reference-free methods
for analyzing sequenced genomes. With large projects sequencing thousands of
individuals, this raises the need for tools capable of handling terabases of
sequence data. A key method is the Burrows-Wheeler transform (BWT), which is
widely used for compressing and indexing reads. We propose a practical
algorithm for building the BWT of a large read collection by merging the BWTs
of subcollections. With our 2.4 Tbp datasets, the algorithm can merge 600
Gbp/day on a single system, using 30 gigabytes of memory overhead on top of the
run-length encoded BWTs.
| [
{
"version": "v1",
"created": "Tue, 3 Nov 2015 13:14:37 GMT"
},
{
"version": "v2",
"created": "Thu, 14 Jan 2016 15:35:19 GMT"
}
] | 2016-01-15T00:00:00 | [
[
"Sirén",
"Jouni",
""
]
] | TITLE: Burrows-Wheeler transform for terabases
ABSTRACT: In order to avoid the reference bias introduced by mapping reads to a
reference genome, bioinformaticians are investigating reference-free methods
for analyzing sequenced genomes. With large projects sequencing thousands of
individuals, this raises the need for tools capable of handling terabases of
sequence data. A key method is the Burrows-Wheeler transform (BWT), which is
widely used for compressing and indexing reads. We propose a practical
algorithm for building the BWT of a large read collection by merging the BWTs
of subcollections. With our 2.4 Tbp datasets, the algorithm can merge 600
Gbp/day on a single system, using 30 gigabytes of memory overhead on top of the
run-length encoded BWTs.
| no_new_dataset | 0.939913 |
1601.03124 | Guangyong Chen | Guangyong Chen, Fengyuan Zhu, Pheng Ann Heng | Online Prediction of Dyadic Data with Heterogeneous Matrix Factorization | 26 pages, 10 figures | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Dyadic Data Prediction (DDP) is an important problem in many research areas.
This paper develops a novel fully Bayesian nonparametric framework which
integrates two popular and complementary approaches, discrete mixed membership
modeling and continuous latent factor modeling into a unified Heterogeneous
Matrix Factorization~(HeMF) model, which can predict the unobserved dyadics
accurately. The HeMF can determine the number of communities automatically and
exploit the latent linear structure for each bicluster efficiently. We propose
a Variational Bayesian method to estimate the parameters and missing data. We
further develop a novel online learning approach for Variational inference and
use it for the online learning of HeMF, which can efficiently cope with the
important large-scale DDP problem. We evaluate the performance of our method on
the EachMoive, MovieLens and Netflix Prize collaborative filtering datasets.
The experiment shows that, our model outperforms state-of-the-art methods on
all benchmarks. Compared with Stochastic Gradient Method (SGD), our online
learning approach achieves significant improvement on the estimation accuracy
and robustness.
| [
{
"version": "v1",
"created": "Wed, 13 Jan 2016 04:20:09 GMT"
}
] | 2016-01-15T00:00:00 | [
[
"Chen",
"Guangyong",
""
],
[
"Zhu",
"Fengyuan",
""
],
[
"Heng",
"Pheng Ann",
""
]
] | TITLE: Online Prediction of Dyadic Data with Heterogeneous Matrix Factorization
ABSTRACT: Dyadic Data Prediction (DDP) is an important problem in many research areas.
This paper develops a novel fully Bayesian nonparametric framework which
integrates two popular and complementary approaches, discrete mixed membership
modeling and continuous latent factor modeling into a unified Heterogeneous
Matrix Factorization~(HeMF) model, which can predict the unobserved dyadics
accurately. The HeMF can determine the number of communities automatically and
exploit the latent linear structure for each bicluster efficiently. We propose
a Variational Bayesian method to estimate the parameters and missing data. We
further develop a novel online learning approach for Variational inference and
use it for the online learning of HeMF, which can efficiently cope with the
important large-scale DDP problem. We evaluate the performance of our method on
the EachMoive, MovieLens and Netflix Prize collaborative filtering datasets.
The experiment shows that, our model outperforms state-of-the-art methods on
all benchmarks. Compared with Stochastic Gradient Method (SGD), our online
learning approach achieves significant improvement on the estimation accuracy
and robustness.
| no_new_dataset | 0.948442 |
1601.03531 | Omar Al-Kadi | O. S. Al-Kadi, Daniel Y.F. Chung, Robert C. Carlisle, Constantin C.
Coussios, J. Alison Noble | Quantification of Ultrasonic Texture heterogeneity via Volumetric
Stochastic Modeling for Tissue Characterization | Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.media.2014.12. 004 | Medical Image Analysis, vol. 21(1), pp. 59-71, 2015 | 10.1016/j.media.2014.12.004 | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Intensity variations in image texture can provide powerful quantitative
information about physical properties of biological tissue. However, tissue
patterns can vary according to the utilized imaging system and are
intrinsically correlated to the scale of analysis. In the case of ultrasound,
the Nakagami distribution is a general model of the ultrasonic backscattering
envelope under various scattering conditions and densities where it can be
employed for characterizing image texture, but the subtle intra-heterogeneities
within a given mass are difficult to capture via this model as it works at a
single spatial scale. This paper proposes a locally adaptive 3D
multi-resolution Nakagami-based fractal feature descriptor that extends
Nakagami-based texture analysis to accommodate subtle speckle spatial frequency
tissue intensity variability in volumetric scans. Local textural fractal
descriptors - which are invariant to affine intensity changes - are extracted
from volumetric patches at different spatial resolutions from voxel
lattice-based generated shape and scale Nakagami parameters. Using ultrasound
radio-frequency datasets we found that after applying an adaptive fractal
decomposition label transfer approach on top of the generated Nakagami voxels,
tissue characterization results were superior to the state of art. Experimental
results on real 3D ultrasonic pre-clinical and clinical datasets suggest that
describing tumor intra-heterogeneity via this descriptor may facilitate
improved prediction of therapy response and disease characterization.
| [
{
"version": "v1",
"created": "Thu, 14 Jan 2016 09:51:37 GMT"
}
] | 2016-01-15T00:00:00 | [
[
"Al-Kadi",
"O. S.",
""
],
[
"Chung",
"Daniel Y. F.",
""
],
[
"Carlisle",
"Robert C.",
""
],
[
"Coussios",
"Constantin C.",
""
],
[
"Noble",
"J. Alison",
""
]
] | TITLE: Quantification of Ultrasonic Texture heterogeneity via Volumetric
Stochastic Modeling for Tissue Characterization
ABSTRACT: Intensity variations in image texture can provide powerful quantitative
information about physical properties of biological tissue. However, tissue
patterns can vary according to the utilized imaging system and are
intrinsically correlated to the scale of analysis. In the case of ultrasound,
the Nakagami distribution is a general model of the ultrasonic backscattering
envelope under various scattering conditions and densities where it can be
employed for characterizing image texture, but the subtle intra-heterogeneities
within a given mass are difficult to capture via this model as it works at a
single spatial scale. This paper proposes a locally adaptive 3D
multi-resolution Nakagami-based fractal feature descriptor that extends
Nakagami-based texture analysis to accommodate subtle speckle spatial frequency
tissue intensity variability in volumetric scans. Local textural fractal
descriptors - which are invariant to affine intensity changes - are extracted
from volumetric patches at different spatial resolutions from voxel
lattice-based generated shape and scale Nakagami parameters. Using ultrasound
radio-frequency datasets we found that after applying an adaptive fractal
decomposition label transfer approach on top of the generated Nakagami voxels,
tissue characterization results were superior to the state of art. Experimental
results on real 3D ultrasonic pre-clinical and clinical datasets suggest that
describing tumor intra-heterogeneity via this descriptor may facilitate
improved prediction of therapy response and disease characterization.
| no_new_dataset | 0.957873 |
1601.03679 | Xiaojun Chang | Xiaojun Chang and Yi Yang and Guodong Long and Chengqi Zhang and
Alexander G. Hauptmann | Dynamic Concept Composition for Zero-Example Event Detection | 7 pages, AAAI 2016 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we focus on automatically detecting events in unconstrained
videos without the use of any visual training exemplars. In principle,
zero-shot learning makes it possible to train an event detection model based on
the assumption that events (e.g. \emph{birthday party}) can be described by
multiple mid-level semantic concepts (e.g. "blowing candle", "birthday cake").
Towards this goal, we first pre-train a bundle of concept classifiers using
data from other sources. Then we evaluate the semantic correlation of each
concept \wrt the event of interest and pick up the relevant concept
classifiers, which are applied on all test videos to get multiple prediction
score vectors. While most existing systems combine the predictions of the
concept classifiers with fixed weights, we propose to learn the optimal weights
of the concept classifiers for each testing video by exploring a set of online
available videos with free-form text descriptions of their content. To validate
the effectiveness of the proposed approach, we have conducted extensive
experiments on the latest TRECVID MEDTest 2014, MEDTest 2013 and CCV dataset.
The experimental results confirm the superiority of the proposed approach.
| [
{
"version": "v1",
"created": "Thu, 14 Jan 2016 17:40:09 GMT"
}
] | 2016-01-15T00:00:00 | [
[
"Chang",
"Xiaojun",
""
],
[
"Yang",
"Yi",
""
],
[
"Long",
"Guodong",
""
],
[
"Zhang",
"Chengqi",
""
],
[
"Hauptmann",
"Alexander G.",
""
]
] | TITLE: Dynamic Concept Composition for Zero-Example Event Detection
ABSTRACT: In this paper, we focus on automatically detecting events in unconstrained
videos without the use of any visual training exemplars. In principle,
zero-shot learning makes it possible to train an event detection model based on
the assumption that events (e.g. \emph{birthday party}) can be described by
multiple mid-level semantic concepts (e.g. "blowing candle", "birthday cake").
Towards this goal, we first pre-train a bundle of concept classifiers using
data from other sources. Then we evaluate the semantic correlation of each
concept \wrt the event of interest and pick up the relevant concept
classifiers, which are applied on all test videos to get multiple prediction
score vectors. While most existing systems combine the predictions of the
concept classifiers with fixed weights, we propose to learn the optimal weights
of the concept classifiers for each testing video by exploring a set of online
available videos with free-form text descriptions of their content. To validate
the effectiveness of the proposed approach, we have conducted extensive
experiments on the latest TRECVID MEDTest 2014, MEDTest 2013 and CCV dataset.
The experimental results confirm the superiority of the proposed approach.
| no_new_dataset | 0.945298 |
1506.08350 | Yadong Mu | Yadong Mu and Wei Liu and Wei Fan | Stochastic Gradient Made Stable: A Manifold Propagation Approach for
Large-Scale Optimization | 14 pages, 9 figures | null | null | null | cs.LG cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Stochastic gradient descent (SGD) holds as a classical method to build large
scale machine learning models over big data. A stochastic gradient is typically
calculated from a limited number of samples (known as mini-batch), so it
potentially incurs a high variance and causes the estimated parameters bounce
around the optimal solution. To improve the stability of stochastic gradient,
recent years have witnessed the proposal of several semi-stochastic gradient
descent algorithms, which distinguish themselves from standard SGD by
incorporating global information into gradient computation. In this paper we
contribute a novel stratified semi-stochastic gradient descent (S3GD) algorithm
to this nascent research area, accelerating the optimization of a large family
of composite convex functions. Though theoretically converging faster, prior
semi-stochastic algorithms are found to suffer from high iteration complexity,
which makes them even slower than SGD in practice on many datasets. In our
proposed S3GD, the semi-stochastic gradient is calculated based on efficient
manifold propagation, which can be numerically accomplished by sparse matrix
multiplications. This way S3GD is able to generate a highly-accurate estimate
of the exact gradient from each mini-batch with largely-reduced computational
complexity. Theoretic analysis reveals that the proposed S3GD elegantly
balances the geometric algorithmic convergence rate against the space and time
complexities during the optimization. The efficacy of S3GD is also
experimentally corroborated on several large-scale benchmark datasets.
| [
{
"version": "v1",
"created": "Sun, 28 Jun 2015 03:33:38 GMT"
},
{
"version": "v2",
"created": "Tue, 12 Jan 2016 21:30:08 GMT"
}
] | 2016-01-14T00:00:00 | [
[
"Mu",
"Yadong",
""
],
[
"Liu",
"Wei",
""
],
[
"Fan",
"Wei",
""
]
] | TITLE: Stochastic Gradient Made Stable: A Manifold Propagation Approach for
Large-Scale Optimization
ABSTRACT: Stochastic gradient descent (SGD) holds as a classical method to build large
scale machine learning models over big data. A stochastic gradient is typically
calculated from a limited number of samples (known as mini-batch), so it
potentially incurs a high variance and causes the estimated parameters bounce
around the optimal solution. To improve the stability of stochastic gradient,
recent years have witnessed the proposal of several semi-stochastic gradient
descent algorithms, which distinguish themselves from standard SGD by
incorporating global information into gradient computation. In this paper we
contribute a novel stratified semi-stochastic gradient descent (S3GD) algorithm
to this nascent research area, accelerating the optimization of a large family
of composite convex functions. Though theoretically converging faster, prior
semi-stochastic algorithms are found to suffer from high iteration complexity,
which makes them even slower than SGD in practice on many datasets. In our
proposed S3GD, the semi-stochastic gradient is calculated based on efficient
manifold propagation, which can be numerically accomplished by sparse matrix
multiplications. This way S3GD is able to generate a highly-accurate estimate
of the exact gradient from each mini-batch with largely-reduced computational
complexity. Theoretic analysis reveals that the proposed S3GD elegantly
balances the geometric algorithmic convergence rate against the space and time
complexities during the optimization. The efficacy of S3GD is also
experimentally corroborated on several large-scale benchmark datasets.
| no_new_dataset | 0.939748 |
1601.02093 | Olivier Mor\`ere | Olivier Mor\`ere, Antoine Veillard, Jie Lin, Julie Petta, Vijay
Chandrasekhar, Tomaso Poggio | Group Invariant Deep Representations for Image Instance Retrieval | null | null | null | null | cs.CV cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most image instance retrieval pipelines are based on comparison of vectors
known as global image descriptors between a query image and the database
images. Due to their success in large scale image classification,
representations extracted from Convolutional Neural Networks (CNN) are quickly
gaining ground on Fisher Vectors (FVs) as state-of-the-art global descriptors
for image instance retrieval. While CNN-based descriptors are generally
remarked for good retrieval performance at lower bitrates, they nevertheless
present a number of drawbacks including the lack of robustness to common object
transformations such as rotations compared with their interest point based FV
counterparts.
In this paper, we propose a method for computing invariant global descriptors
from CNNs. Our method implements a recently proposed mathematical theory for
invariance in a sensory cortex modeled as a feedforward neural network. The
resulting global descriptors can be made invariant to multiple arbitrary
transformation groups while retaining good discriminativeness.
Based on a thorough empirical evaluation using several publicly available
datasets, we show that our method is able to significantly and consistently
improve retrieval results every time a new type of invariance is incorporated.
We also show that our method which has few parameters is not prone to
overfitting: improvements generalize well across datasets with different
properties with regard to invariances. Finally, we show that our descriptors
are able to compare favourably to other state-of-the-art compact descriptors in
similar bitranges, exceeding the highest retrieval results reported in the
literature on some datasets. A dedicated dimensionality reduction step
--quantization or hashing-- may be able to further improve the competitiveness
of the descriptors.
| [
{
"version": "v1",
"created": "Sat, 9 Jan 2016 10:42:35 GMT"
},
{
"version": "v2",
"created": "Wed, 13 Jan 2016 06:43:44 GMT"
}
] | 2016-01-14T00:00:00 | [
[
"Morère",
"Olivier",
""
],
[
"Veillard",
"Antoine",
""
],
[
"Lin",
"Jie",
""
],
[
"Petta",
"Julie",
""
],
[
"Chandrasekhar",
"Vijay",
""
],
[
"Poggio",
"Tomaso",
""
]
] | TITLE: Group Invariant Deep Representations for Image Instance Retrieval
ABSTRACT: Most image instance retrieval pipelines are based on comparison of vectors
known as global image descriptors between a query image and the database
images. Due to their success in large scale image classification,
representations extracted from Convolutional Neural Networks (CNN) are quickly
gaining ground on Fisher Vectors (FVs) as state-of-the-art global descriptors
for image instance retrieval. While CNN-based descriptors are generally
remarked for good retrieval performance at lower bitrates, they nevertheless
present a number of drawbacks including the lack of robustness to common object
transformations such as rotations compared with their interest point based FV
counterparts.
In this paper, we propose a method for computing invariant global descriptors
from CNNs. Our method implements a recently proposed mathematical theory for
invariance in a sensory cortex modeled as a feedforward neural network. The
resulting global descriptors can be made invariant to multiple arbitrary
transformation groups while retaining good discriminativeness.
Based on a thorough empirical evaluation using several publicly available
datasets, we show that our method is able to significantly and consistently
improve retrieval results every time a new type of invariance is incorporated.
We also show that our method which has few parameters is not prone to
overfitting: improvements generalize well across datasets with different
properties with regard to invariances. Finally, we show that our descriptors
are able to compare favourably to other state-of-the-art compact descriptors in
similar bitranges, exceeding the highest retrieval results reported in the
literature on some datasets. A dedicated dimensionality reduction step
--quantization or hashing-- may be able to further improve the competitiveness
of the descriptors.
| no_new_dataset | 0.948155 |
1601.03295 | Gabriela Csurka | Gabriela Csurka | Document image classification, with a specific view on applications of
patent images | Paper submitted in 2014 as book chapter of Current Challenges in
Patent Information Retrieval, Second edition by M. Lupu et al (eds.). To
appear in 2016 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The main focus of this paper is document image classification and retrieval,
where we analyze and compare different parameters for the RunLeght Histogram
(RL) and Fisher Vector (FV) based image representations. We do an exhaustive
experimental study using different document image datasets, including the MARG
benchmarks, two datasets built on customer data and the images from the Patent
Image Classification task of the Clef-IP 2011. The aim of the study is to give
guidelines on how to best choose the parameters such that the same features
perform well on different tasks. As an example of such need, we describe the
Image-based Patent Retrieval task's of Clef-IP 2011, where we used the same
image representation to predict the image type and retrieve relevant patents.
| [
{
"version": "v1",
"created": "Wed, 13 Jan 2016 16:02:13 GMT"
}
] | 2016-01-14T00:00:00 | [
[
"Csurka",
"Gabriela",
""
]
] | TITLE: Document image classification, with a specific view on applications of
patent images
ABSTRACT: The main focus of this paper is document image classification and retrieval,
where we analyze and compare different parameters for the RunLeght Histogram
(RL) and Fisher Vector (FV) based image representations. We do an exhaustive
experimental study using different document image datasets, including the MARG
benchmarks, two datasets built on customer data and the images from the Patent
Image Classification task of the Clef-IP 2011. The aim of the study is to give
guidelines on how to best choose the parameters such that the same features
perform well on different tasks. As an example of such need, we describe the
Image-based Patent Retrieval task's of Clef-IP 2011, where we used the same
image representation to predict the image type and retrieve relevant patents.
| no_new_dataset | 0.945197 |
1601.03354 | Alexey Grigorev | Alexey Grigorev | Identifier Namespaces in Mathematical Notation | Master Thesis defended at TU Berlin in Summer 2015 | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this thesis, we look at the problem of assigning each identifier of a
document to a namespace. At the moment, there does not exist a special dataset
where all identifiers are grouped to namespaces, and therefore we need to
create such a dataset ourselves.
To do that, we need to find groups of documents that use identifiers in the
same way. This can be done with cluster analysis methods. We argue that
documents can be represented by the identifiers they contain, and this approach
is similar to representing textual information in the Vector Space Model.
Because of this, we can apply traditional document clustering techniques for
namespace discovery.
Because the problem is new, there is no gold standard dataset, and it is hard
to evaluate the performance of our method. To overcome it, we first use Java
source code as a dataset for our experiments, since it contains the namespace
information. We verify that our method can partially recover namespaces from
source code using only information about identifiers.
The algorithms are evaluated on the English Wikipedia, and the proposed
method can extract namespaces on a variety of topics. After extraction, the
namespaces are organized into a hierarchical structure by using existing
classification schemes such as MSC, PACS and ACM. We also apply it to the
Russian Wikipedia, and the results are consistent across the languages.
To our knowledge, the problem of introducing namespaces to mathematics has
not been studied before, and prior to our work there has been no dataset where
identifiers are grouped into namespaces. Thus, our result is not only a good
start, but also a good indicator that automatic namespace discovery is
possible.
| [
{
"version": "v1",
"created": "Wed, 13 Jan 2016 19:17:00 GMT"
}
] | 2016-01-14T00:00:00 | [
[
"Grigorev",
"Alexey",
""
]
] | TITLE: Identifier Namespaces in Mathematical Notation
ABSTRACT: In this thesis, we look at the problem of assigning each identifier of a
document to a namespace. At the moment, there does not exist a special dataset
where all identifiers are grouped to namespaces, and therefore we need to
create such a dataset ourselves.
To do that, we need to find groups of documents that use identifiers in the
same way. This can be done with cluster analysis methods. We argue that
documents can be represented by the identifiers they contain, and this approach
is similar to representing textual information in the Vector Space Model.
Because of this, we can apply traditional document clustering techniques for
namespace discovery.
Because the problem is new, there is no gold standard dataset, and it is hard
to evaluate the performance of our method. To overcome it, we first use Java
source code as a dataset for our experiments, since it contains the namespace
information. We verify that our method can partially recover namespaces from
source code using only information about identifiers.
The algorithms are evaluated on the English Wikipedia, and the proposed
method can extract namespaces on a variety of topics. After extraction, the
namespaces are organized into a hierarchical structure by using existing
classification schemes such as MSC, PACS and ACM. We also apply it to the
Russian Wikipedia, and the results are consistent across the languages.
To our knowledge, the problem of introducing namespaces to mathematics has
not been studied before, and prior to our work there has been no dataset where
identifiers are grouped into namespaces. Thus, our result is not only a good
start, but also a good indicator that automatic namespace discovery is
possible.
| new_dataset | 0.973519 |
1504.03763 | Facundo Memoli | Tamal K. Dey, Facundo Memoli, Yusu Wang | Mutiscale Mapper: A Framework for Topological Summarization of Data and
Maps | null | null | null | null | cs.CG math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Summarizing topological information from datasets and maps defined on them is
a central theme in topological data analysis. \textsf{Mapper}, a tool for such
summarization, takes as input both a possibly high dimensional dataset and a
map defined on the data, and produces a summary of the data by using a cover of
the codomain of the map. This cover, via a pullback operation to the domain,
produces a simplicial complex connecting the data points.
The resulting view of the data through a cover of the codomain offers
flexibility in analyzing the data. However, it offers only a view at a fixed
scale at which the cover is constructed. Inspired by the concept, we explore a
notion of a tower of covers which induces a tower of simplicial complexes
connected by simplicial maps, which we call {\em multiscale mapper}. We study
the resulting structure, its stability, and design practical algorithms to
compute its associated persistence diagrams efficiently. Specifically, when the
domain is a simplicial complex and the map is a real-valued piecewise-linear
function, the algorithm can compute the exact persistence diagram only from the
1-skeleton of the input complex. For general maps, we present a combinatorial
version of the algorithm that acts only on \emph{vertex sets} connected by the
1-skeleton graph, and this algorithm approximates the exact persistence diagram
thanks to a stability result that we show to hold. We also relate the
multiscale mapper with the \v{C}ech complexes arising from a natural pullback
pseudometric defined on the input domain.
| [
{
"version": "v1",
"created": "Wed, 15 Apr 2015 01:47:21 GMT"
},
{
"version": "v2",
"created": "Tue, 12 Jan 2016 16:28:20 GMT"
}
] | 2016-01-13T00:00:00 | [
[
"Dey",
"Tamal K.",
""
],
[
"Memoli",
"Facundo",
""
],
[
"Wang",
"Yusu",
""
]
] | TITLE: Mutiscale Mapper: A Framework for Topological Summarization of Data and
Maps
ABSTRACT: Summarizing topological information from datasets and maps defined on them is
a central theme in topological data analysis. \textsf{Mapper}, a tool for such
summarization, takes as input both a possibly high dimensional dataset and a
map defined on the data, and produces a summary of the data by using a cover of
the codomain of the map. This cover, via a pullback operation to the domain,
produces a simplicial complex connecting the data points.
The resulting view of the data through a cover of the codomain offers
flexibility in analyzing the data. However, it offers only a view at a fixed
scale at which the cover is constructed. Inspired by the concept, we explore a
notion of a tower of covers which induces a tower of simplicial complexes
connected by simplicial maps, which we call {\em multiscale mapper}. We study
the resulting structure, its stability, and design practical algorithms to
compute its associated persistence diagrams efficiently. Specifically, when the
domain is a simplicial complex and the map is a real-valued piecewise-linear
function, the algorithm can compute the exact persistence diagram only from the
1-skeleton of the input complex. For general maps, we present a combinatorial
version of the algorithm that acts only on \emph{vertex sets} connected by the
1-skeleton graph, and this algorithm approximates the exact persistence diagram
thanks to a stability result that we show to hold. We also relate the
multiscale mapper with the \v{C}ech complexes arising from a natural pullback
pseudometric defined on the input domain.
| no_new_dataset | 0.949529 |
1512.04077 | Mojmir Mutny | Mojmir Mutny, Rahul Nair and Jens-Malte Gottfried | Learning the Correction for Multi-Path Deviations in Time-of-Flight
Cameras | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Multipath effect in Time-of-Flight(ToF) cameras still remains to be a
challenging problem that hinders further processing of 3D data information.
Based on the evidence from previous literature, we explored the possibility of
using machine learning techniques to correct this effect. Firstly, we created
two new datasets of of ToF images rendered via ToF simulator of LuxRender.
These two datasets contain corners in multiple orientations and with different
material properties. We chose scenes with corners as multipath effects are most
pronounced in corners. Secondly, we used this dataset to construct a learning
model to predict real valued corrections to the ToF data using Random Forests.
We found out that in our smaller dataset we were able to predict real valued
correction and improve the quality of depth images significantly by removing
multipath bias. With our algorithm, we improved relative per-pixel error from
average value of 19% to 3%. Additionally, variance of the error was lowered by
an order of magnitude.
| [
{
"version": "v1",
"created": "Sun, 13 Dec 2015 16:31:14 GMT"
},
{
"version": "v2",
"created": "Tue, 12 Jan 2016 11:17:58 GMT"
}
] | 2016-01-13T00:00:00 | [
[
"Mutny",
"Mojmir",
""
],
[
"Nair",
"Rahul",
""
],
[
"Gottfried",
"Jens-Malte",
""
]
] | TITLE: Learning the Correction for Multi-Path Deviations in Time-of-Flight
Cameras
ABSTRACT: The Multipath effect in Time-of-Flight(ToF) cameras still remains to be a
challenging problem that hinders further processing of 3D data information.
Based on the evidence from previous literature, we explored the possibility of
using machine learning techniques to correct this effect. Firstly, we created
two new datasets of of ToF images rendered via ToF simulator of LuxRender.
These two datasets contain corners in multiple orientations and with different
material properties. We chose scenes with corners as multipath effects are most
pronounced in corners. Secondly, we used this dataset to construct a learning
model to predict real valued corrections to the ToF data using Random Forests.
We found out that in our smaller dataset we were able to predict real valued
correction and improve the quality of depth images significantly by removing
multipath bias. With our algorithm, we improved relative per-pixel error from
average value of 19% to 3%. Additionally, variance of the error was lowered by
an order of magnitude.
| new_dataset | 0.961642 |
1512.05685 | Johann Schaible | Johann Schaible and Thomas Gottron and Ansgar Scherp | TermPicker: Enabling the Reuse of Vocabulary Terms by Exploiting Data
from the Linked Open Data Cloud - An Extended Technical Report | 17 pages, 3 figures, extended technical report for a Conference Paper | null | null | null | cs.DB cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deciding which vocabulary terms to use when modeling data as Linked Open Data
(LOD) is far from trivial. Choosing too general vocabulary terms, or terms from
vocabularies that are not used by other LOD datasets, is likely to lead to a
data representation, which will be harder to understand by humans and to be
consumed by Linked data applications. In this technical report, we propose
TermPicker: a novel approach for vocabulary reuse by recommending RDF types and
properties based on exploiting the information on how other data providers on
the LOD cloud use RDF types and properties to describe their data. To this end,
we introduce the notion of so-called schema-level patterns (SLPs). They capture
how sets of RDF types are connected via sets of properties within some data
collection, e.g., within a dataset on the LOD cloud. TermPicker uses such SLPs
and generates a ranked list of vocabulary terms for reuse. The lists of
recommended terms are ordered by a ranking model which is computed using the
machine learning approach Learning To Rank (L2R). TermPicker is evaluated based
on the recommendation quality that is measured using the Mean Average Precision
(MAP) and the Mean Reciprocal Rank at the first five positions (MRR@5). Our
results illustrate an improvement of the recommendation quality by 29% - 36%
when using SLPs compared to the beforehand investigated baselines of
recommending solely popular vocabulary terms or terms from the same vocabulary.
The overall best results are achieved using SLPs in conjunction with the
Learning To Rank algorithm Random Forests.
| [
{
"version": "v1",
"created": "Thu, 17 Dec 2015 17:37:56 GMT"
},
{
"version": "v2",
"created": "Mon, 11 Jan 2016 22:00:10 GMT"
}
] | 2016-01-13T00:00:00 | [
[
"Schaible",
"Johann",
""
],
[
"Gottron",
"Thomas",
""
],
[
"Scherp",
"Ansgar",
""
]
] | TITLE: TermPicker: Enabling the Reuse of Vocabulary Terms by Exploiting Data
from the Linked Open Data Cloud - An Extended Technical Report
ABSTRACT: Deciding which vocabulary terms to use when modeling data as Linked Open Data
(LOD) is far from trivial. Choosing too general vocabulary terms, or terms from
vocabularies that are not used by other LOD datasets, is likely to lead to a
data representation, which will be harder to understand by humans and to be
consumed by Linked data applications. In this technical report, we propose
TermPicker: a novel approach for vocabulary reuse by recommending RDF types and
properties based on exploiting the information on how other data providers on
the LOD cloud use RDF types and properties to describe their data. To this end,
we introduce the notion of so-called schema-level patterns (SLPs). They capture
how sets of RDF types are connected via sets of properties within some data
collection, e.g., within a dataset on the LOD cloud. TermPicker uses such SLPs
and generates a ranked list of vocabulary terms for reuse. The lists of
recommended terms are ordered by a ranking model which is computed using the
machine learning approach Learning To Rank (L2R). TermPicker is evaluated based
on the recommendation quality that is measured using the Mean Average Precision
(MAP) and the Mean Reciprocal Rank at the first five positions (MRR@5). Our
results illustrate an improvement of the recommendation quality by 29% - 36%
when using SLPs compared to the beforehand investigated baselines of
recommending solely popular vocabulary terms or terms from the same vocabulary.
The overall best results are achieved using SLPs in conjunction with the
Learning To Rank algorithm Random Forests.
| no_new_dataset | 0.955361 |
1601.02603 | Marian-Andrei Rizoiu | Marian-Andrei Rizoiu, Julien Velcin, St\'ephane Lallich | How to Use Temporal-Driven Constrained Clustering to Detect Typical
Evolutions | null | Int. J. Artif. Intell. Tools 23, 1460013 (2014) [26 pages] | 10.1142/S0218213014600136 | null | cs.LG cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose a new time-aware dissimilarity measure that takes
into account the temporal dimension. Observations that are close in the
description space, but distant in time are considered as dissimilar. We also
propose a method to enforce the segmentation contiguity, by introducing, in the
objective function, a penalty term inspired from the Normal Distribution
Function. We combine the two propositions into a novel time-driven constrained
clustering algorithm, called TDCK-Means, which creates a partition of coherent
clusters, both in the multidimensional space and in the temporal space. This
algorithm uses soft semi-supervised constraints, to encourage adjacent
observations belonging to the same entity to be assigned to the same cluster.
We apply our algorithm to a Political Studies dataset in order to detect
typical evolution phases. We adapt the Shannon entropy in order to measure the
entity contiguity, and we show that our proposition consistently improves
temporal cohesion of clusters, without any significant loss in the
multidimensional variance.
| [
{
"version": "v1",
"created": "Mon, 11 Jan 2016 01:20:26 GMT"
}
] | 2016-01-13T00:00:00 | [
[
"Rizoiu",
"Marian-Andrei",
""
],
[
"Velcin",
"Julien",
""
],
[
"Lallich",
"Stéphane",
""
]
] | TITLE: How to Use Temporal-Driven Constrained Clustering to Detect Typical
Evolutions
ABSTRACT: In this paper, we propose a new time-aware dissimilarity measure that takes
into account the temporal dimension. Observations that are close in the
description space, but distant in time are considered as dissimilar. We also
propose a method to enforce the segmentation contiguity, by introducing, in the
objective function, a penalty term inspired from the Normal Distribution
Function. We combine the two propositions into a novel time-driven constrained
clustering algorithm, called TDCK-Means, which creates a partition of coherent
clusters, both in the multidimensional space and in the temporal space. This
algorithm uses soft semi-supervised constraints, to encourage adjacent
observations belonging to the same entity to be assigned to the same cluster.
We apply our algorithm to a Political Studies dataset in order to detect
typical evolution phases. We adapt the Shannon entropy in order to measure the
entity contiguity, and we show that our proposition consistently improves
temporal cohesion of clusters, without any significant loss in the
multidimensional variance.
| no_new_dataset | 0.95018 |
1601.02705 | Jaeyong Sung | Jaeyong Sung, Seok Hyun Jin, Ian Lenz, Ashutosh Saxena | Robobarista: Learning to Manipulate Novel Objects via Deep Multimodal
Embedding | Journal Version | null | null | null | cs.RO cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | There is a large variety of objects and appliances in human environments,
such as stoves, coffee dispensers, juice extractors, and so on. It is
challenging for a roboticist to program a robot for each of these object types
and for each of their instantiations. In this work, we present a novel approach
to manipulation planning based on the idea that many household objects share
similarly-operated object parts. We formulate the manipulation planning as a
structured prediction problem and learn to transfer manipulation strategy
across different objects by embedding point-cloud, natural language, and
manipulation trajectory data into a shared embedding space using a deep neural
network. In order to learn semantically meaningful spaces throughout our
network, we introduce a method for pre-training its lower layers for multimodal
feature embedding and a method for fine-tuning this embedding space using a
loss-based margin. In order to collect a large number of manipulation
demonstrations for different objects, we develop a new crowd-sourcing platform
called Robobarista. We test our model on our dataset consisting of 116 objects
and appliances with 249 parts along with 250 language instructions, for which
there are 1225 crowd-sourced manipulation demonstrations. We further show that
our robot with our model can even prepare a cup of a latte with appliances it
has never seen before.
| [
{
"version": "v1",
"created": "Tue, 12 Jan 2016 00:56:30 GMT"
}
] | 2016-01-13T00:00:00 | [
[
"Sung",
"Jaeyong",
""
],
[
"Jin",
"Seok Hyun",
""
],
[
"Lenz",
"Ian",
""
],
[
"Saxena",
"Ashutosh",
""
]
] | TITLE: Robobarista: Learning to Manipulate Novel Objects via Deep Multimodal
Embedding
ABSTRACT: There is a large variety of objects and appliances in human environments,
such as stoves, coffee dispensers, juice extractors, and so on. It is
challenging for a roboticist to program a robot for each of these object types
and for each of their instantiations. In this work, we present a novel approach
to manipulation planning based on the idea that many household objects share
similarly-operated object parts. We formulate the manipulation planning as a
structured prediction problem and learn to transfer manipulation strategy
across different objects by embedding point-cloud, natural language, and
manipulation trajectory data into a shared embedding space using a deep neural
network. In order to learn semantically meaningful spaces throughout our
network, we introduce a method for pre-training its lower layers for multimodal
feature embedding and a method for fine-tuning this embedding space using a
loss-based margin. In order to collect a large number of manipulation
demonstrations for different objects, we develop a new crowd-sourcing platform
called Robobarista. We test our model on our dataset consisting of 116 objects
and appliances with 249 parts along with 250 language instructions, for which
there are 1225 crowd-sourced manipulation demonstrations. We further show that
our robot with our model can even prepare a cup of a latte with appliances it
has never seen before.
| new_dataset | 0.957078 |
1601.02733 | Ehsan Hosseini-Asl | Ehsan Hosseini-Asl, Jacek M. Zurada, Olfa Nasraoui | Deep Learning of Part-based Representation of Data Using Sparse
Autoencoders with Nonnegativity Constraints | Accepted for publication in IEEE Transactions of Neural Networks and
Learning Systems | null | 10.1109/TNNLS.2015.2479223 | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We demonstrate a new deep learning autoencoder network, trained by a
nonnegativity constraint algorithm (NCAE), that learns features which show
part-based representation of data. The learning algorithm is based on
constraining negative weights. The performance of the algorithm is assessed
based on decomposing data into parts and its prediction performance is tested
on three standard image data sets and one text dataset. The results indicate
that the nonnegativity constraint forces the autoencoder to learn features that
amount to a part-based representation of data, while improving sparsity and
reconstruction quality in comparison with the traditional sparse autoencoder
and Nonnegative Matrix Factorization. It is also shown that this newly acquired
representation improves the prediction performance of a deep neural network.
| [
{
"version": "v1",
"created": "Tue, 12 Jan 2016 05:33:03 GMT"
}
] | 2016-01-13T00:00:00 | [
[
"Hosseini-Asl",
"Ehsan",
""
],
[
"Zurada",
"Jacek M.",
""
],
[
"Nasraoui",
"Olfa",
""
]
] | TITLE: Deep Learning of Part-based Representation of Data Using Sparse
Autoencoders with Nonnegativity Constraints
ABSTRACT: We demonstrate a new deep learning autoencoder network, trained by a
nonnegativity constraint algorithm (NCAE), that learns features which show
part-based representation of data. The learning algorithm is based on
constraining negative weights. The performance of the algorithm is assessed
based on decomposing data into parts and its prediction performance is tested
on three standard image data sets and one text dataset. The results indicate
that the nonnegativity constraint forces the autoencoder to learn features that
amount to a part-based representation of data, while improving sparsity and
reconstruction quality in comparison with the traditional sparse autoencoder
and Nonnegative Matrix Factorization. It is also shown that this newly acquired
representation improves the prediction performance of a deep neural network.
| no_new_dataset | 0.947137 |
1411.6520 | Ilya Trofimov | Ilya Trofimov, Alexander Genkin | Distributed Coordinate Descent for L1-regularized Logistic Regression | null | Analysis of Images, Social Networks and Texts. Fourth
International Conference, AIST 2015, Yekaterinburg, Russia, April 9-11, 2015,
Revised Selected Papers. Communications in Computer and Information Science,
Vol. 542, 243-254, Springer | 10.1007/978-3-319-26123-2_24 | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Solving logistic regression with L1-regularization in distributed settings is
an important problem. This problem arises when training dataset is very large
and cannot fit the memory of a single machine. We present d-GLMNET, a new
algorithm solving logistic regression with L1-regularization in the distributed
settings. We empirically show that it is superior over distributed online
learning via truncated gradient.
| [
{
"version": "v1",
"created": "Mon, 24 Nov 2014 16:40:33 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Trofimov",
"Ilya",
""
],
[
"Genkin",
"Alexander",
""
]
] | TITLE: Distributed Coordinate Descent for L1-regularized Logistic Regression
ABSTRACT: Solving logistic regression with L1-regularization in distributed settings is
an important problem. This problem arises when training dataset is very large
and cannot fit the memory of a single machine. We present d-GLMNET, a new
algorithm solving logistic regression with L1-regularization in the distributed
settings. We empirically show that it is superior over distributed online
learning via truncated gradient.
| no_new_dataset | 0.94743 |
1503.01313 | Matej Kristan | Matej Kristan, Jiri Matas, Ales Leonardis, Tomas Vojir, Roman
Pflugfelder, Gustavo Fernandez, Georg Nebehay, Fatih Porikli and Luka Cehovin | A Novel Performance Evaluation Methodology for Single-Target Trackers | Final version (Accepted), IEEE Pattern Analysis and Machine
Intelligence, 2016 | null | 10.1109/TPAMI.2016.2516982 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper addresses the problem of single-target tracker performance
evaluation. We consider the performance measures, the dataset and the
evaluation system to be the most important components of tracker evaluation and
propose requirements for each of them. The requirements are the basis of a new
evaluation methodology that aims at a simple and easily interpretable tracker
comparison. The ranking-based methodology addresses tracker equivalence in
terms of statistical significance and practical differences. A fully-annotated
dataset with per-frame annotations with several visual attributes is
introduced. The diversity of its visual properties is maximized in a novel way
by clustering a large number of videos according to their visual attributes.
This makes it the most sophistically constructed and annotated dataset to date.
A multi-platform evaluation system allowing easy integration of third-party
trackers is presented as well. The proposed evaluation methodology was tested
on the VOT2014 challenge on the new dataset and 38 trackers, making it the
largest benchmark to date. Most of the tested trackers are indeed
state-of-the-art since they outperform the standard baselines, resulting in a
highly-challenging benchmark. An exhaustive analysis of the dataset from the
perspective of tracking difficulty is carried out. To facilitate tracker
comparison a new performance visualization technique is proposed.
| [
{
"version": "v1",
"created": "Wed, 4 Mar 2015 14:12:17 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Apr 2015 14:00:23 GMT"
},
{
"version": "v3",
"created": "Fri, 8 Jan 2016 15:27:11 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Kristan",
"Matej",
""
],
[
"Matas",
"Jiri",
""
],
[
"Leonardis",
"Ales",
""
],
[
"Vojir",
"Tomas",
""
],
[
"Pflugfelder",
"Roman",
""
],
[
"Fernandez",
"Gustavo",
""
],
[
"Nebehay",
"Georg",
""
],
[
"Porikli",
"Fatih",
""
],
[
"Cehovin",
"Luka",
""
]
] | TITLE: A Novel Performance Evaluation Methodology for Single-Target Trackers
ABSTRACT: This paper addresses the problem of single-target tracker performance
evaluation. We consider the performance measures, the dataset and the
evaluation system to be the most important components of tracker evaluation and
propose requirements for each of them. The requirements are the basis of a new
evaluation methodology that aims at a simple and easily interpretable tracker
comparison. The ranking-based methodology addresses tracker equivalence in
terms of statistical significance and practical differences. A fully-annotated
dataset with per-frame annotations with several visual attributes is
introduced. The diversity of its visual properties is maximized in a novel way
by clustering a large number of videos according to their visual attributes.
This makes it the most sophistically constructed and annotated dataset to date.
A multi-platform evaluation system allowing easy integration of third-party
trackers is presented as well. The proposed evaluation methodology was tested
on the VOT2014 challenge on the new dataset and 38 trackers, making it the
largest benchmark to date. Most of the tested trackers are indeed
state-of-the-art since they outperform the standard baselines, resulting in a
highly-challenging benchmark. An exhaustive analysis of the dataset from the
perspective of tracking difficulty is carried out. To facilitate tracker
comparison a new performance visualization technique is proposed.
| new_dataset | 0.959383 |
1504.03655 | Yingyu Liang | Bo Xie, Yingyu Liang, Le Song | Scale Up Nonlinear Component Analysis with Doubly Stochastic Gradients | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nonlinear component analysis such as kernel Principle Component Analysis
(KPCA) and kernel Canonical Correlation Analysis (KCCA) are widely used in
machine learning, statistics and data analysis, but they can not scale up to
big datasets. Recent attempts have employed random feature approximations to
convert the problem to the primal form for linear computational complexity.
However, to obtain high quality solutions, the number of random features should
be the same order of magnitude as the number of data points, making such
approach not directly applicable to the regime with millions of data points.
We propose a simple, computationally efficient, and memory friendly algorithm
based on the "doubly stochastic gradients" to scale up a range of kernel
nonlinear component analysis, such as kernel PCA, CCA and SVD. Despite the
\emph{non-convex} nature of these problems, our method enjoys theoretical
guarantees that it converges at the rate $\tilde{O}(1/t)$ to the global
optimum, even for the top $k$ eigen subspace. Unlike many alternatives, our
algorithm does not require explicit orthogonalization, which is infeasible on
big datasets. We demonstrate the effectiveness and scalability of our algorithm
on large scale synthetic and real world datasets.
| [
{
"version": "v1",
"created": "Tue, 14 Apr 2015 18:34:03 GMT"
},
{
"version": "v2",
"created": "Tue, 23 Jun 2015 02:47:45 GMT"
},
{
"version": "v3",
"created": "Sun, 12 Jul 2015 23:09:21 GMT"
},
{
"version": "v4",
"created": "Sun, 10 Jan 2016 22:54:59 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Xie",
"Bo",
""
],
[
"Liang",
"Yingyu",
""
],
[
"Song",
"Le",
""
]
] | TITLE: Scale Up Nonlinear Component Analysis with Doubly Stochastic Gradients
ABSTRACT: Nonlinear component analysis such as kernel Principle Component Analysis
(KPCA) and kernel Canonical Correlation Analysis (KCCA) are widely used in
machine learning, statistics and data analysis, but they can not scale up to
big datasets. Recent attempts have employed random feature approximations to
convert the problem to the primal form for linear computational complexity.
However, to obtain high quality solutions, the number of random features should
be the same order of magnitude as the number of data points, making such
approach not directly applicable to the regime with millions of data points.
We propose a simple, computationally efficient, and memory friendly algorithm
based on the "doubly stochastic gradients" to scale up a range of kernel
nonlinear component analysis, such as kernel PCA, CCA and SVD. Despite the
\emph{non-convex} nature of these problems, our method enjoys theoretical
guarantees that it converges at the rate $\tilde{O}(1/t)$ to the global
optimum, even for the top $k$ eigen subspace. Unlike many alternatives, our
algorithm does not require explicit orthogonalization, which is infeasible on
big datasets. We demonstrate the effectiveness and scalability of our algorithm
on large scale synthetic and real world datasets.
| no_new_dataset | 0.948632 |
1507.05409 | Bhaskar Mukhoty | Bhaskar Mukhoty, Ruchir Gupta and Y. N. Singh | A Parameter-free Affinity Based Clustering | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Several methods have been proposed to estimate the number of clusters in a
dataset; the basic ideal behind all of them has been to study an index that
measures inter-cluster separation and intra-cluster cohesion over a range of
cluster numbers and report the number which gives an optimum value of the
index. In this paper we propose a simple, parameter free approach that is like
human cognition to form clusters, where closely lying points are easily
identified to form a cluster and total number of clusters are revealed. To
identify closely lying points, affinity of two points is defined as a function
of distance and a threshold affinity is identified, above which two points in a
dataset are likely to be in the same cluster. Well separated clusters are
identified even in the presence of outliers, whereas for not so well separated
dataset, final number of clusters are estimated and the detected clusters are
merged to produce the final clusters. Experiments performed with several large
dimensional synthetic and real datasets show good results with robustness to
noise and density variation within dataset.
| [
{
"version": "v1",
"created": "Mon, 20 Jul 2015 07:59:17 GMT"
},
{
"version": "v2",
"created": "Mon, 11 Jan 2016 10:24:38 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Mukhoty",
"Bhaskar",
""
],
[
"Gupta",
"Ruchir",
""
],
[
"Singh",
"Y. N.",
""
]
] | TITLE: A Parameter-free Affinity Based Clustering
ABSTRACT: Several methods have been proposed to estimate the number of clusters in a
dataset; the basic ideal behind all of them has been to study an index that
measures inter-cluster separation and intra-cluster cohesion over a range of
cluster numbers and report the number which gives an optimum value of the
index. In this paper we propose a simple, parameter free approach that is like
human cognition to form clusters, where closely lying points are easily
identified to form a cluster and total number of clusters are revealed. To
identify closely lying points, affinity of two points is defined as a function
of distance and a threshold affinity is identified, above which two points in a
dataset are likely to be in the same cluster. Well separated clusters are
identified even in the presence of outliers, whereas for not so well separated
dataset, final number of clusters are estimated and the detected clusters are
merged to produce the final clusters. Experiments performed with several large
dimensional synthetic and real datasets show good results with robustness to
noise and density variation within dataset.
| no_new_dataset | 0.95803 |
1508.03110 | Michael B Hynes | Manda Winlaw, Michael B. Hynes, Anthony Caterini, Hans De Sterck | Algorithmic Acceleration of Parallel ALS for Collaborative Filtering:
Speeding up Distributed Big Data Recommendation in Spark | Proceedings of ICPADS 2015, Melbourne, AU. 10 pages; 6 figures; 4
tables | null | null | null | math.NA cs.DC cs.IR cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Collaborative filtering algorithms are important building blocks in many
practical recommendation systems. For example, many large-scale data processing
environments include collaborative filtering models for which the Alternating
Least Squares (ALS) algorithm is used to compute latent factor matrix
decompositions. In this paper, we propose an approach to accelerate the
convergence of parallel ALS-based optimization methods for collaborative
filtering using a nonlinear conjugate gradient (NCG) wrapper around the ALS
iterations. We also provide a parallel implementation of the accelerated
ALS-NCG algorithm in the Apache Spark distributed data processing environment,
and an efficient line search technique as part of the ALS-NCG implementation
that requires only one pass over the data on distributed datasets. In serial
numerical experiments on a linux workstation and parallel numerical experiments
on a 16 node cluster with 256 computing cores, we demonstrate that the combined
ALS-NCG method requires many fewer iterations and less time than standalone ALS
to reach movie rankings with high accuracy on the MovieLens 20M dataset. In
parallel, ALS-NCG can achieve an acceleration factor of 4 or greater in clock
time when an accurate solution is desired; furthermore, the acceleration factor
increases as greater numerical precision is required in the solution. In
addition, the NCG acceleration mechanism is efficient in parallel and scales
linearly with problem size on synthetic datasets with up to nearly 1 billion
ratings. The acceleration mechanism is general and may also be applicable to
other optimization methods for collaborative filtering.
| [
{
"version": "v1",
"created": "Thu, 13 Aug 2015 03:37:04 GMT"
},
{
"version": "v2",
"created": "Wed, 28 Oct 2015 16:53:49 GMT"
},
{
"version": "v3",
"created": "Sun, 10 Jan 2016 23:52:03 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Winlaw",
"Manda",
""
],
[
"Hynes",
"Michael B.",
""
],
[
"Caterini",
"Anthony",
""
],
[
"De Sterck",
"Hans",
""
]
] | TITLE: Algorithmic Acceleration of Parallel ALS for Collaborative Filtering:
Speeding up Distributed Big Data Recommendation in Spark
ABSTRACT: Collaborative filtering algorithms are important building blocks in many
practical recommendation systems. For example, many large-scale data processing
environments include collaborative filtering models for which the Alternating
Least Squares (ALS) algorithm is used to compute latent factor matrix
decompositions. In this paper, we propose an approach to accelerate the
convergence of parallel ALS-based optimization methods for collaborative
filtering using a nonlinear conjugate gradient (NCG) wrapper around the ALS
iterations. We also provide a parallel implementation of the accelerated
ALS-NCG algorithm in the Apache Spark distributed data processing environment,
and an efficient line search technique as part of the ALS-NCG implementation
that requires only one pass over the data on distributed datasets. In serial
numerical experiments on a linux workstation and parallel numerical experiments
on a 16 node cluster with 256 computing cores, we demonstrate that the combined
ALS-NCG method requires many fewer iterations and less time than standalone ALS
to reach movie rankings with high accuracy on the MovieLens 20M dataset. In
parallel, ALS-NCG can achieve an acceleration factor of 4 or greater in clock
time when an accurate solution is desired; furthermore, the acceleration factor
increases as greater numerical precision is required in the solution. In
addition, the NCG acceleration mechanism is efficient in parallel and scales
linearly with problem size on synthetic datasets with up to nearly 1 billion
ratings. The acceleration mechanism is general and may also be applicable to
other optimization methods for collaborative filtering.
| no_new_dataset | 0.951459 |
1509.00838 | Hongyuan Mei | Hongyuan Mei and Mohit Bansal and Matthew R. Walter | What to talk about and how? Selective Generation using LSTMs with
Coarse-to-Fine Alignment | null | null | null | null | cs.CL cs.AI cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose an end-to-end, domain-independent neural encoder-aligner-decoder
model for selective generation, i.e., the joint task of content selection and
surface realization. Our model first encodes a full set of over-determined
database event records via an LSTM-based recurrent neural network, then
utilizes a novel coarse-to-fine aligner to identify the small subset of salient
records to talk about, and finally employs a decoder to generate free-form
descriptions of the aligned, selected records. Our model achieves the best
selection and generation results reported to-date (with 59% relative
improvement in generation) on the benchmark WeatherGov dataset, despite using
no specialized features or linguistic resources. Using an improved k-nearest
neighbor beam filter helps further. We also perform a series of ablations and
visualizations to elucidate the contributions of our key model components.
Lastly, we evaluate the generalizability of our model on the RoboCup dataset,
and get results that are competitive with or better than the state-of-the-art,
despite being severely data-starved.
| [
{
"version": "v1",
"created": "Wed, 2 Sep 2015 19:52:56 GMT"
},
{
"version": "v2",
"created": "Fri, 8 Jan 2016 23:07:32 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Mei",
"Hongyuan",
""
],
[
"Bansal",
"Mohit",
""
],
[
"Walter",
"Matthew R.",
""
]
] | TITLE: What to talk about and how? Selective Generation using LSTMs with
Coarse-to-Fine Alignment
ABSTRACT: We propose an end-to-end, domain-independent neural encoder-aligner-decoder
model for selective generation, i.e., the joint task of content selection and
surface realization. Our model first encodes a full set of over-determined
database event records via an LSTM-based recurrent neural network, then
utilizes a novel coarse-to-fine aligner to identify the small subset of salient
records to talk about, and finally employs a decoder to generate free-form
descriptions of the aligned, selected records. Our model achieves the best
selection and generation results reported to-date (with 59% relative
improvement in generation) on the benchmark WeatherGov dataset, despite using
no specialized features or linguistic resources. Using an improved k-nearest
neighbor beam filter helps further. We also perform a series of ablations and
visualizations to elucidate the contributions of our key model components.
Lastly, we evaluate the generalizability of our model on the RoboCup dataset,
and get results that are competitive with or better than the state-of-the-art,
despite being severely data-starved.
| no_new_dataset | 0.947137 |
1511.07394 | Kevin Shih | Kevin J. Shih, Saurabh Singh, Derek Hoiem | Where To Look: Focus Regions for Visual Question Answering | Submitted to CVPR2016 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a method that learns to answer visual questions by selecting image
regions relevant to the text-based query. Our method exhibits significant
improvements in answering questions such as "what color," where it is necessary
to evaluate a specific location, and "what room," where it selectively
identifies informative image regions. Our model is tested on the VQA dataset
which is the largest human-annotated visual question answering dataset to our
knowledge.
| [
{
"version": "v1",
"created": "Mon, 23 Nov 2015 20:17:18 GMT"
},
{
"version": "v2",
"created": "Sun, 10 Jan 2016 13:26:23 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Shih",
"Kevin J.",
""
],
[
"Singh",
"Saurabh",
""
],
[
"Hoiem",
"Derek",
""
]
] | TITLE: Where To Look: Focus Regions for Visual Question Answering
ABSTRACT: We present a method that learns to answer visual questions by selecting image
regions relevant to the text-based query. Our method exhibits significant
improvements in answering questions such as "what color," where it is necessary
to evaluate a specific location, and "what room," where it selectively
identifies informative image regions. Our model is tested on the VQA dataset
which is the largest human-annotated visual question answering dataset to our
knowledge.
| new_dataset | 0.833663 |
1601.02034 | Ayush Jain | Ayush Jain, Joon Young Seo, Karan Goel, Andrew Kuznetsov, Aditya
Parameswaran, Hari Sundaram | It's just a matter of perspective(s): Crowd-Powered Consensus
Organization of Corpora | null | null | null | null | cs.DB cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the problem of organizing a collection of objects - images, videos -
into clusters, using crowdsourcing. This problem is notoriously hard for
computers to do automatically, and even with crowd workers, is challenging to
orchestrate: (a) workers may cluster based on different latent hierarchies or
perspectives; (b) workers may cluster at different granularities even when
clustering using the same perspective; and (c) workers may only see a small
portion of the objects when deciding how to cluster them (and therefore have
limited understanding of the "big picture"). We develop cost-efficient,
accurate algorithms for identifying the consensus organization (i.e., the
organizing perspective most workers prefer to employ), and incorporate these
algorithms into a cost-effective workflow for organizing a collection of
objects, termed ORCHESTRA. We compare our algorithms with other algorithms for
clustering, on a variety of real-world datasets, and demonstrate that ORCHESTRA
organizes items better and at significantly lower costs.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2016 21:31:56 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Jain",
"Ayush",
""
],
[
"Seo",
"Joon Young",
""
],
[
"Goel",
"Karan",
""
],
[
"Kuznetsov",
"Andrew",
""
],
[
"Parameswaran",
"Aditya",
""
],
[
"Sundaram",
"Hari",
""
]
] | TITLE: It's just a matter of perspective(s): Crowd-Powered Consensus
Organization of Corpora
ABSTRACT: We study the problem of organizing a collection of objects - images, videos -
into clusters, using crowdsourcing. This problem is notoriously hard for
computers to do automatically, and even with crowd workers, is challenging to
orchestrate: (a) workers may cluster based on different latent hierarchies or
perspectives; (b) workers may cluster at different granularities even when
clustering using the same perspective; and (c) workers may only see a small
portion of the objects when deciding how to cluster them (and therefore have
limited understanding of the "big picture"). We develop cost-efficient,
accurate algorithms for identifying the consensus organization (i.e., the
organizing perspective most workers prefer to employ), and incorporate these
algorithms into a cost-effective workflow for organizing a collection of
objects, termed ORCHESTRA. We compare our algorithms with other algorithms for
clustering, on a variety of real-world datasets, and demonstrate that ORCHESTRA
organizes items better and at significantly lower costs.
| no_new_dataset | 0.953232 |
1601.02047 | Panagiotis Tolias | P. Tolias | Low energy electron reflection from tungsten surfaces | 4 pages, 4 figures | null | null | null | physics.plasm-ph cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The incidence of very low energy electrons on metal surfaces is mainly
dictated by the phenomenon of quantum mechanical reflection at the metal
interface. Low energy electron reflection is insignificant in higher energy
regimes, where the more familiar secondary electron emission and electron
backscattering processes are the dominant features of the electron-metal
interaction. It is a highly controversial subject that has mostly emerged
during the last years. In this brief note we examine the source of the
controversy, present some basic theoretical considerations, recommend a dataset
of reliable experimental results for the reflection of low energy electrons
from tungsten surfaces and discuss the suppression of reflected electrons by
external magnetic fields in the light of applications in fusion devices.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2016 22:50:39 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Tolias",
"P.",
""
]
] | TITLE: Low energy electron reflection from tungsten surfaces
ABSTRACT: The incidence of very low energy electrons on metal surfaces is mainly
dictated by the phenomenon of quantum mechanical reflection at the metal
interface. Low energy electron reflection is insignificant in higher energy
regimes, where the more familiar secondary electron emission and electron
backscattering processes are the dominant features of the electron-metal
interaction. It is a highly controversial subject that has mostly emerged
during the last years. In this brief note we examine the source of the
controversy, present some basic theoretical considerations, recommend a dataset
of reliable experimental results for the reflection of low energy electrons
from tungsten surfaces and discuss the suppression of reflected electrons by
external magnetic fields in the light of applications in fusion devices.
| no_new_dataset | 0.908496 |
1601.02071 | Eduardo Graells-Garrido | Eduardo Graells-Garrido, Mounia Lalmas, Ricardo Baeza-Yates | Sentiment Visualisation Widgets for Exploratory Search | Presented at the Social Personalization Workshop held jointly with
ACM Hypertext 2014. 6 pages | null | null | null | cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes the usage of \emph{visualisation widgets} for exploratory
search with \emph{sentiment} as a facet. Starting from specific design goals
for depiction of ambivalence in sentiment, two visualization widgets were
implemented: \emph{scatter plot} and \emph{parallel coordinates}. Those widgets
were evaluated against a text baseline in a small-scale usability study with
exploratory tasks using Wikipedia as dataset. The study results indicate that
users spend more time browsing with scatter plots in a positive way. A post-hoc
analysis of individual differences in behavior revealed that when considering
two types of users, \emph{explorers} and \emph{achievers}, engagement with
scatter plots is positive and significantly greater \textit{when users are
explorers}. We discuss the implications of these findings for sentiment-based
exploratory search and personalised user interfaces.
| [
{
"version": "v1",
"created": "Sat, 9 Jan 2016 03:48:07 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Graells-Garrido",
"Eduardo",
""
],
[
"Lalmas",
"Mounia",
""
],
[
"Baeza-Yates",
"Ricardo",
""
]
] | TITLE: Sentiment Visualisation Widgets for Exploratory Search
ABSTRACT: This paper proposes the usage of \emph{visualisation widgets} for exploratory
search with \emph{sentiment} as a facet. Starting from specific design goals
for depiction of ambivalence in sentiment, two visualization widgets were
implemented: \emph{scatter plot} and \emph{parallel coordinates}. Those widgets
were evaluated against a text baseline in a small-scale usability study with
exploratory tasks using Wikipedia as dataset. The study results indicate that
users spend more time browsing with scatter plots in a positive way. A post-hoc
analysis of individual differences in behavior revealed that when considering
two types of users, \emph{explorers} and \emph{achievers}, engagement with
scatter plots is positive and significantly greater \textit{when users are
explorers}. We discuss the implications of these findings for sentiment-based
exploratory search and personalised user interfaces.
| no_new_dataset | 0.957118 |
1601.02197 | Wei-Long Zheng | Wei-Long Zheng, Jia-Yi Zhu, Bao-Liang Lu | Identifying Stable Patterns over Time for Emotion Recognition from EEG | null | null | null | null | cs.HC cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we investigate stable patterns of electroencephalogram (EEG)
over time for emotion recognition using a machine learning approach. Up to now,
various findings of activated patterns associated with different emotions have
been reported. However, their stability over time has not been fully
investigated yet. In this paper, we focus on identifying EEG stability in
emotion recognition. To validate the efficiency of the machine learning
algorithms used in this study, we systematically evaluate the performance of
various popular feature extraction, feature selection, feature smoothing and
pattern classification methods with the DEAP dataset and a newly developed
dataset for this study. The experimental results indicate that stable patterns
exhibit consistency across sessions; the lateral temporal areas activate more
for positive emotion than negative one in beta and gamma bands; the neural
patterns of neutral emotion have higher alpha responses at parietal and
occipital sites; and for negative emotion, the neural patterns have significant
higher delta responses at parietal and occipital sites and higher gamma
responses at prefrontal sites. The performance of our emotion recognition
system shows that the neural patterns are relatively stable within and between
sessions.
| [
{
"version": "v1",
"created": "Sun, 10 Jan 2016 10:43:24 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Zheng",
"Wei-Long",
""
],
[
"Zhu",
"Jia-Yi",
""
],
[
"Lu",
"Bao-Liang",
""
]
] | TITLE: Identifying Stable Patterns over Time for Emotion Recognition from EEG
ABSTRACT: In this paper, we investigate stable patterns of electroencephalogram (EEG)
over time for emotion recognition using a machine learning approach. Up to now,
various findings of activated patterns associated with different emotions have
been reported. However, their stability over time has not been fully
investigated yet. In this paper, we focus on identifying EEG stability in
emotion recognition. To validate the efficiency of the machine learning
algorithms used in this study, we systematically evaluate the performance of
various popular feature extraction, feature selection, feature smoothing and
pattern classification methods with the DEAP dataset and a newly developed
dataset for this study. The experimental results indicate that stable patterns
exhibit consistency across sessions; the lateral temporal areas activate more
for positive emotion than negative one in beta and gamma bands; the neural
patterns of neutral emotion have higher alpha responses at parietal and
occipital sites; and for negative emotion, the neural patterns have significant
higher delta responses at parietal and occipital sites and higher gamma
responses at prefrontal sites. The performance of our emotion recognition
system shows that the neural patterns are relatively stable within and between
sessions.
| new_dataset | 0.959687 |
1601.02220 | Anurag Arnab | Anurag Arnab, Michael Sapienza, Stuart Golodetz, Julien Valentin,
Ondrej Miksik, Shahram Izadi, Philip Torr | Joint Object-Material Category Segmentation from Audio-Visual Cues | Published in British Machine Vision Conference (BMVC) 2015 | null | null | null | cs.CV cs.SD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is not always possible to recognise objects and infer material properties
for a scene from visual cues alone, since objects can look visually similar
whilst being made of very different materials. In this paper, we therefore
present an approach that augments the available dense visual cues with sparse
auditory cues in order to estimate dense object and material labels. Since
estimates of object class and material properties are mutually informative, we
optimise our multi-output labelling jointly using a random-field framework. We
evaluate our system on a new dataset with paired visual and auditory data that
we make publicly available. We demonstrate that this joint estimation of object
and material labels significantly outperforms the estimation of either category
in isolation.
| [
{
"version": "v1",
"created": "Sun, 10 Jan 2016 14:14:53 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Arnab",
"Anurag",
""
],
[
"Sapienza",
"Michael",
""
],
[
"Golodetz",
"Stuart",
""
],
[
"Valentin",
"Julien",
""
],
[
"Miksik",
"Ondrej",
""
],
[
"Izadi",
"Shahram",
""
],
[
"Torr",
"Philip",
""
]
] | TITLE: Joint Object-Material Category Segmentation from Audio-Visual Cues
ABSTRACT: It is not always possible to recognise objects and infer material properties
for a scene from visual cues alone, since objects can look visually similar
whilst being made of very different materials. In this paper, we therefore
present an approach that augments the available dense visual cues with sparse
auditory cues in order to estimate dense object and material labels. Since
estimates of object class and material properties are mutually informative, we
optimise our multi-output labelling jointly using a random-field framework. We
evaluate our system on a new dataset with paired visual and auditory data that
we make publicly available. We demonstrate that this joint estimation of object
and material labels significantly outperforms the estimation of either category
in isolation.
| new_dataset | 0.959724 |
1601.02487 | Ahmed Bassiouny | Abubakrelsedik Karali, Ahmad Bassiouny and Motaz El-Saban | Facial Expression Recognition in the Wild using Rich Deep Features | in International Conference in Image Processing, 2015 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Facial Expression Recognition is an active area of research in computer
vision with a wide range of applications. Several approaches have been
developed to solve this problem for different benchmark datasets. However,
Facial Expression Recognition in the wild remains an area where much work is
still needed to serve real-world applications. To this end, in this paper we
present a novel approach towards facial expression recognition. We fuse rich
deep features with domain knowledge through encoding discriminant facial
patches. We conduct experiments on two of the most popular benchmark datasets;
CK and TFE. Moreover, we present a novel dataset that, unlike its precedents,
consists of natural - not acted - expression images. Experimental results show
that our approach achieves state-of-the-art results over standard benchmarks
and our own dataset
| [
{
"version": "v1",
"created": "Mon, 11 Jan 2016 15:52:27 GMT"
}
] | 2016-01-12T00:00:00 | [
[
"Karali",
"Abubakrelsedik",
""
],
[
"Bassiouny",
"Ahmad",
""
],
[
"El-Saban",
"Motaz",
""
]
] | TITLE: Facial Expression Recognition in the Wild using Rich Deep Features
ABSTRACT: Facial Expression Recognition is an active area of research in computer
vision with a wide range of applications. Several approaches have been
developed to solve this problem for different benchmark datasets. However,
Facial Expression Recognition in the wild remains an area where much work is
still needed to serve real-world applications. To this end, in this paper we
present a novel approach towards facial expression recognition. We fuse rich
deep features with domain knowledge through encoding discriminant facial
patches. We conduct experiments on two of the most popular benchmark datasets;
CK and TFE. Moreover, we present a novel dataset that, unlike its precedents,
consists of natural - not acted - expression images. Experimental results show
that our approach achieves state-of-the-art results over standard benchmarks
and our own dataset
| new_dataset | 0.961207 |
1511.06328 | Shuangfei Zhai | Shuangfei Zhai, Zhongfei Zhang | Manifold Regularized Discriminative Neural Networks | In submission to ICLR 2016 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Unregularized deep neural networks (DNNs) can be easily overfit with a
limited sample size. We argue that this is mostly due to the disriminative
nature of DNNs which directly model the conditional probability (or score) of
labels given the input. The ignorance of input distribution makes DNNs
difficult to generalize to unseen data. Recent advances in regularization
techniques, such as pretraining and dropout, indicate that modeling input data
distribution (either explicitly or implicitly) greatly improves the
generalization ability of a DNN. In this work, we explore the manifold
hypothesis which assumes that instances within the same class lie in a smooth
manifold. We accordingly propose two simple regularizers to a standard
discriminative DNN. The first one, named Label-Aware Manifold Regularization,
assumes the availability of labels and penalizes large norms of the loss
function w.r.t. data points. The second one, named Label-Independent Manifold
Regularization, does not use label information and instead penalizes the
Frobenius norm of the Jacobian matrix of prediction scores w.r.t. data points,
which makes semi-supervised learning possible. We perform extensive control
experiments on fully supervised and semi-supervised tasks using the MNIST,
CIFAR10 and SVHN datasets and achieve excellent results.
| [
{
"version": "v1",
"created": "Thu, 19 Nov 2015 19:46:39 GMT"
},
{
"version": "v2",
"created": "Thu, 3 Dec 2015 17:11:25 GMT"
},
{
"version": "v3",
"created": "Thu, 7 Jan 2016 22:05:56 GMT"
}
] | 2016-01-11T00:00:00 | [
[
"Zhai",
"Shuangfei",
""
],
[
"Zhang",
"Zhongfei",
""
]
] | TITLE: Manifold Regularized Discriminative Neural Networks
ABSTRACT: Unregularized deep neural networks (DNNs) can be easily overfit with a
limited sample size. We argue that this is mostly due to the disriminative
nature of DNNs which directly model the conditional probability (or score) of
labels given the input. The ignorance of input distribution makes DNNs
difficult to generalize to unseen data. Recent advances in regularization
techniques, such as pretraining and dropout, indicate that modeling input data
distribution (either explicitly or implicitly) greatly improves the
generalization ability of a DNN. In this work, we explore the manifold
hypothesis which assumes that instances within the same class lie in a smooth
manifold. We accordingly propose two simple regularizers to a standard
discriminative DNN. The first one, named Label-Aware Manifold Regularization,
assumes the availability of labels and penalizes large norms of the loss
function w.r.t. data points. The second one, named Label-Independent Manifold
Regularization, does not use label information and instead penalizes the
Frobenius norm of the Jacobian matrix of prediction scores w.r.t. data points,
which makes semi-supervised learning possible. We perform extensive control
experiments on fully supervised and semi-supervised tasks using the MNIST,
CIFAR10 and SVHN datasets and achieve excellent results.
| no_new_dataset | 0.949809 |
1511.06434 | Alec Radford | Alec Radford, Luke Metz, and Soumith Chintala | Unsupervised Representation Learning with Deep Convolutional Generative
Adversarial Networks | Under review as a conference paper at ICLR 2016 | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, supervised learning with convolutional networks (CNNs) has
seen huge adoption in computer vision applications. Comparatively, unsupervised
learning with CNNs has received less attention. In this work we hope to help
bridge the gap between the success of CNNs for supervised learning and
unsupervised learning. We introduce a class of CNNs called deep convolutional
generative adversarial networks (DCGANs), that have certain architectural
constraints, and demonstrate that they are a strong candidate for unsupervised
learning. Training on various image datasets, we show convincing evidence that
our deep convolutional adversarial pair learns a hierarchy of representations
from object parts to scenes in both the generator and discriminator.
Additionally, we use the learned features for novel tasks - demonstrating their
applicability as general image representations.
| [
{
"version": "v1",
"created": "Thu, 19 Nov 2015 22:50:32 GMT"
},
{
"version": "v2",
"created": "Thu, 7 Jan 2016 23:09:39 GMT"
}
] | 2016-01-11T00:00:00 | [
[
"Radford",
"Alec",
""
],
[
"Metz",
"Luke",
""
],
[
"Chintala",
"Soumith",
""
]
] | TITLE: Unsupervised Representation Learning with Deep Convolutional Generative
Adversarial Networks
ABSTRACT: In recent years, supervised learning with convolutional networks (CNNs) has
seen huge adoption in computer vision applications. Comparatively, unsupervised
learning with CNNs has received less attention. In this work we hope to help
bridge the gap between the success of CNNs for supervised learning and
unsupervised learning. We introduce a class of CNNs called deep convolutional
generative adversarial networks (DCGANs), that have certain architectural
constraints, and demonstrate that they are a strong candidate for unsupervised
learning. Training on various image datasets, we show convincing evidence that
our deep convolutional adversarial pair learns a hierarchy of representations
from object parts to scenes in both the generator and discriminator.
Additionally, we use the learned features for novel tasks - demonstrating their
applicability as general image representations.
| no_new_dataset | 0.943764 |
1601.01770 | Albert Haque | Albert Haque | A MapReduce Approach to NoSQL RDF Databases | Undergraduate Honors Thesis, December 2013, The University of Texas
at Austin, Department of Computer Science. Report# HR-13-13 (honors theses) | null | null | HR-13-13 | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, the increased need to house and process large volumes of
data has prompted the need for distributed storage and querying systems. The
growth of machine-readable RDF triples has prompted both industry and academia
to develop new database systems, called NoSQL, with characteristics that differ
from classical databases. Many of these systems compromise ACID properties for
increased horizontal scalability and data availability. This thesis concerns
the development and evaluation of a NoSQL triplestore. Triplestores are
database management systems central to emerging technologies such as the
Semantic Web and linked data. The evaluation spans several benchmarks,
including the two most commonly used in triplestore evaluation, the Berlin
SPARQL Benchmark, and the DBpedia benchmark, a query workload that operates an
RDF representation of Wikipedia. Results reveal that the join algorithm used by
the system plays a critical role in dictating query runtimes. Distributed graph
databases must carefully optimize queries before generating MapReduce query
plans as network traffic for large datasets can become prohibitive if the query
is executed naively.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2016 05:04:26 GMT"
}
] | 2016-01-11T00:00:00 | [
[
"Haque",
"Albert",
""
]
] | TITLE: A MapReduce Approach to NoSQL RDF Databases
ABSTRACT: In recent years, the increased need to house and process large volumes of
data has prompted the need for distributed storage and querying systems. The
growth of machine-readable RDF triples has prompted both industry and academia
to develop new database systems, called NoSQL, with characteristics that differ
from classical databases. Many of these systems compromise ACID properties for
increased horizontal scalability and data availability. This thesis concerns
the development and evaluation of a NoSQL triplestore. Triplestores are
database management systems central to emerging technologies such as the
Semantic Web and linked data. The evaluation spans several benchmarks,
including the two most commonly used in triplestore evaluation, the Berlin
SPARQL Benchmark, and the DBpedia benchmark, a query workload that operates an
RDF representation of Wikipedia. Results reveal that the join algorithm used by
the system plays a critical role in dictating query runtimes. Distributed graph
databases must carefully optimize queries before generating MapReduce query
plans as network traffic for large datasets can become prohibitive if the query
is executed naively.
| no_new_dataset | 0.940463 |
1601.01876 | Salah Eddine Bekhouche SE. Bekhouche | Salah Eddine Bekhouche (1), Abdelkrim Ouafi (1), Abdelmalik
Taleb-Ahmed (2), Abdenour Hadid (3), Azeddine Benlamoudi (1) ((1) Laboratory
of LESIA, University of Biskra, Algeria, (2) LAMIH, University of
Valenciennes, France, (3) Center for Machine Vision Research, University of
Oulu, Finland) | Facial age estimation using BSIF and LBP | 5 pages, 8 figures | null | 10.13140/RG.2.1.1933.6483/1 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Human face aging is irreversible process causing changes in human face
characteristics such us hair whitening, muscles drop and wrinkles. Due to the
importance of human face aging in biometrics systems, age estimation became an
attractive area for researchers. This paper presents a novel method to estimate
the age from face images, using binarized statistical image features (BSIF) and
local binary patterns (LBP)histograms as features performed by support vector
regression (SVR) and kernel ridge regression (KRR). We applied our method on
FG-NET and PAL datasets. Our proposed method has shown superiority to that of
the state-of-the-art methods when using the whole PAL database.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2016 14:03:21 GMT"
}
] | 2016-01-11T00:00:00 | [
[
"Bekhouche",
"Salah Eddine",
""
],
[
"Ouafi",
"Abdelkrim",
""
],
[
"Taleb-Ahmed",
"Abdelmalik",
""
],
[
"Hadid",
"Abdenour",
""
],
[
"Benlamoudi",
"Azeddine",
""
]
] | TITLE: Facial age estimation using BSIF and LBP
ABSTRACT: Human face aging is irreversible process causing changes in human face
characteristics such us hair whitening, muscles drop and wrinkles. Due to the
importance of human face aging in biometrics systems, age estimation became an
attractive area for researchers. This paper presents a novel method to estimate
the age from face images, using binarized statistical image features (BSIF) and
local binary patterns (LBP)histograms as features performed by support vector
regression (SVR) and kernel ridge regression (KRR). We applied our method on
FG-NET and PAL datasets. Our proposed method has shown superiority to that of
the state-of-the-art methods when using the whole PAL database.
| no_new_dataset | 0.941815 |
1601.01885 | Anguelos Nicolaou | Anguelos Nicolaou, Andrew Bagdanov, Lluis Gomez-Bigorda, Dimosthenis
Karatzas | Visual Script and Language Identification | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we introduce a script identification method based on
hand-crafted texture features and an artificial neural network. The proposed
pipeline achieves near state-of-the-art performance for script identification
of video-text and state-of-the-art performance on visual language
identification of handwritten text. More than using the deep network as a
classifier, the use of its intermediary activations as a learned metric
demonstrates remarkable results and allows the use of discriminative models on
unknown classes. Comparative experiments in video-text and text in the wild
datasets provide insights on the internals of the proposed deep network.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2016 14:25:20 GMT"
}
] | 2016-01-11T00:00:00 | [
[
"Nicolaou",
"Anguelos",
""
],
[
"Bagdanov",
"Andrew",
""
],
[
"Gomez-Bigorda",
"Lluis",
""
],
[
"Karatzas",
"Dimosthenis",
""
]
] | TITLE: Visual Script and Language Identification
ABSTRACT: In this paper we introduce a script identification method based on
hand-crafted texture features and an artificial neural network. The proposed
pipeline achieves near state-of-the-art performance for script identification
of video-text and state-of-the-art performance on visual language
identification of handwritten text. More than using the deep network as a
classifier, the use of its intermediary activations as a learned metric
demonstrates remarkable results and allows the use of discriminative models on
unknown classes. Comparative experiments in video-text and text in the wild
datasets provide insights on the internals of the proposed deep network.
| no_new_dataset | 0.947672 |
1601.01940 | Giuseppe Jurman | Giuseppe Jurman | Metric projection for dynamic multiplex networks | null | null | null | null | physics.soc-ph math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Evolving multiplex networks are a powerful model for representing the
dynamics along time of different phenomena, such as social networks, power
grids, biological pathways. However, exploring the structure of the multiplex
network time series is still an open problem. Here we propose a two-steps
strategy to tackle this problem based on the concept of distance (metric)
between networks. Given a multiplex graph, first a network of networks is built
for each time steps, and then a real valued time series is obtained by the
sequence of (simple) networks by evaluating the distance from the first element
of the series. The effectiveness of this approach in detecting the occurring
changes along the original time series is shown on a synthetic example first,
and then on the Gulf dataset of political events.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2016 16:50:14 GMT"
}
] | 2016-01-11T00:00:00 | [
[
"Jurman",
"Giuseppe",
""
]
] | TITLE: Metric projection for dynamic multiplex networks
ABSTRACT: Evolving multiplex networks are a powerful model for representing the
dynamics along time of different phenomena, such as social networks, power
grids, biological pathways. However, exploring the structure of the multiplex
network time series is still an open problem. Here we propose a two-steps
strategy to tackle this problem based on the concept of distance (metric)
between networks. Given a multiplex graph, first a network of networks is built
for each time steps, and then a real valued time series is obtained by the
sequence of (simple) networks by evaluating the distance from the first element
of the series. The effectiveness of this approach in detecting the occurring
changes along the original time series is shown on a synthetic example first,
and then on the Gulf dataset of political events.
| no_new_dataset | 0.953405 |
1412.3750 | Jeremy Debattista | Jeremy Debattista, Christoph Lange, S\"oren Auer | Luzzu - A Framework for Linked Data Quality Assessment | null | null | null | null | cs.DB cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the increasing adoption and growth of the Linked Open Data cloud [9],
with RDFa, Microformats and other ways of embedding data into ordinary Web
pages, and with initiatives such as schema.org, the Web is currently being
complemented with a Web of Data. Thus, the Web of Data shares many
characteristics with the original Web of Documents, which also varies in
quality. This heterogeneity makes it challenging to determine the quality of
the data published on the Web and to subsequently make this information
explicit to data consumers. The main contribution of this article is LUZZU, a
quality assessment framework for Linked Open Data. Apart from providing quality
metadata and quality problem reports that can be used for data cleaning, LUZZU
is extensible: third party metrics can be easily plugged-in the framework. The
framework does not rely on SPARQL endpoints, and is thus free of all the
problems that come with them, such as query timeouts. Another advantage over
SPARQL based qual- ity assessment frameworks is that metrics implemented in
LUZZU can have more complex functionality than triple matching. Using the
framework, we performed a quality assessment of a number of statistical linked
datasets that are available on the LOD cloud. For this evaluation, 25 metrics
from ten different dimensions were implemented.
| [
{
"version": "v1",
"created": "Thu, 11 Dec 2014 18:28:47 GMT"
},
{
"version": "v2",
"created": "Tue, 5 May 2015 15:01:16 GMT"
},
{
"version": "v3",
"created": "Thu, 7 Jan 2016 17:19:41 GMT"
}
] | 2016-01-08T00:00:00 | [
[
"Debattista",
"Jeremy",
""
],
[
"Lange",
"Christoph",
""
],
[
"Auer",
"Sören",
""
]
] | TITLE: Luzzu - A Framework for Linked Data Quality Assessment
ABSTRACT: With the increasing adoption and growth of the Linked Open Data cloud [9],
with RDFa, Microformats and other ways of embedding data into ordinary Web
pages, and with initiatives such as schema.org, the Web is currently being
complemented with a Web of Data. Thus, the Web of Data shares many
characteristics with the original Web of Documents, which also varies in
quality. This heterogeneity makes it challenging to determine the quality of
the data published on the Web and to subsequently make this information
explicit to data consumers. The main contribution of this article is LUZZU, a
quality assessment framework for Linked Open Data. Apart from providing quality
metadata and quality problem reports that can be used for data cleaning, LUZZU
is extensible: third party metrics can be easily plugged-in the framework. The
framework does not rely on SPARQL endpoints, and is thus free of all the
problems that come with them, such as query timeouts. Another advantage over
SPARQL based qual- ity assessment frameworks is that metrics implemented in
LUZZU can have more complex functionality than triple matching. Using the
framework, we performed a quality assessment of a number of statistical linked
datasets that are available on the LOD cloud. For this evaluation, 25 metrics
from ten different dimensions were implemented.
| no_new_dataset | 0.944022 |
1505.05007 | Paul Blomstedt PhD | Paul Blomstedt, Ritabrata Dutta, Sohan Seth, Alvis Brazma and Samuel
Kaski | Modelling-based experiment retrieval: A case study with gene expression
clustering | Updated figures. The final version of this article will appear in
Bioinformatics (https://bioinformatics.oxfordjournals.org/) | null | 10.1093/bioinformatics/btv762 | null | stat.ML cs.IR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Motivation: Public and private repositories of experimental data are growing
to sizes that require dedicated methods for finding relevant data. To improve
on the state of the art of keyword searches from annotations, methods for
content-based retrieval have been proposed. In the context of gene expression
experiments, most methods retrieve gene expression profiles, requiring each
experiment to be expressed as a single profile, typically of case vs. control.
A more general, recently suggested alternative is to retrieve experiments whose
models are good for modelling the query dataset. However, for very noisy and
high-dimensional query data, this retrieval criterion turns out to be very
noisy as well.
Results: We propose doing retrieval using a denoised model of the query
dataset, instead of the original noisy dataset itself. To this end, we
introduce a general probabilistic framework, where each experiment is modelled
separately and the retrieval is done by finding related models. For retrieval
of gene expression experiments, we use a probabilistic model called product
partition model, which induces a clustering of genes that show similar
expression patterns across a number of samples. The suggested metric for
retrieval using clusterings is the normalized information distance. Empirical
results finally suggest that inference for the full probabilistic model can be
approximated with good performance using computationally faster heuristic
clustering approaches (e.g. $k$-means). The method is highly scalable and
straightforward to apply to construct a general-purpose gene expression
experiment retrieval method.
Availability: The method can be implemented using standard clustering
algorithms and normalized information distance, available in many statistical
software packages.
| [
{
"version": "v1",
"created": "Tue, 19 May 2015 14:21:34 GMT"
},
{
"version": "v2",
"created": "Tue, 26 May 2015 11:53:47 GMT"
},
{
"version": "v3",
"created": "Mon, 23 Nov 2015 09:12:58 GMT"
},
{
"version": "v4",
"created": "Mon, 4 Jan 2016 15:08:26 GMT"
}
] | 2016-01-08T00:00:00 | [
[
"Blomstedt",
"Paul",
""
],
[
"Dutta",
"Ritabrata",
""
],
[
"Seth",
"Sohan",
""
],
[
"Brazma",
"Alvis",
""
],
[
"Kaski",
"Samuel",
""
]
] | TITLE: Modelling-based experiment retrieval: A case study with gene expression
clustering
ABSTRACT: Motivation: Public and private repositories of experimental data are growing
to sizes that require dedicated methods for finding relevant data. To improve
on the state of the art of keyword searches from annotations, methods for
content-based retrieval have been proposed. In the context of gene expression
experiments, most methods retrieve gene expression profiles, requiring each
experiment to be expressed as a single profile, typically of case vs. control.
A more general, recently suggested alternative is to retrieve experiments whose
models are good for modelling the query dataset. However, for very noisy and
high-dimensional query data, this retrieval criterion turns out to be very
noisy as well.
Results: We propose doing retrieval using a denoised model of the query
dataset, instead of the original noisy dataset itself. To this end, we
introduce a general probabilistic framework, where each experiment is modelled
separately and the retrieval is done by finding related models. For retrieval
of gene expression experiments, we use a probabilistic model called product
partition model, which induces a clustering of genes that show similar
expression patterns across a number of samples. The suggested metric for
retrieval using clusterings is the normalized information distance. Empirical
results finally suggest that inference for the full probabilistic model can be
approximated with good performance using computationally faster heuristic
clustering approaches (e.g. $k$-means). The method is highly scalable and
straightforward to apply to construct a general-purpose gene expression
experiment retrieval method.
Availability: The method can be implemented using standard clustering
algorithms and normalized information distance, available in many statistical
software packages.
| no_new_dataset | 0.950227 |
1511.04103 | Panqu Wang | Panqu Wang, Garrison W. Cottrell | Basic Level Categorization Facilitates Visual Object Recognition | ICLR 2016 submission R1 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in deep learning have led to significant progress in the
computer vision field, especially for visual object recognition tasks. The
features useful for object classification are learned by feed-forward deep
convolutional neural networks (CNNs) automatically, and they are shown to be
able to predict and decode neural representations in the ventral visual pathway
of humans and monkeys. However, despite the huge amount of work on optimizing
CNNs, there has not been much research focused on linking CNNs with guiding
principles from the human visual cortex. In this work, we propose a network
optimization strategy inspired by both of the developmental trajectory of
children's visual object recognition capabilities, and Bar (2003), who
hypothesized that basic level information is carried in the fast magnocellular
pathway through the prefrontal cortex (PFC) and then projected back to inferior
temporal cortex (IT), where subordinate level categorization is achieved. We
instantiate this idea by training a deep CNN to perform basic level object
categorization first, and then train it on subordinate level categorization. We
apply this idea to training AlexNet (Krizhevsky et al., 2012) on the ILSVRC
2012 dataset and show that the top-5 accuracy increases from 80.13% to 82.14%,
demonstrating the effectiveness of the method. We also show that subsequent
transfer learning on smaller datasets gives superior results.
| [
{
"version": "v1",
"created": "Thu, 12 Nov 2015 21:41:35 GMT"
},
{
"version": "v2",
"created": "Thu, 19 Nov 2015 21:47:35 GMT"
},
{
"version": "v3",
"created": "Thu, 7 Jan 2016 08:26:54 GMT"
}
] | 2016-01-08T00:00:00 | [
[
"Wang",
"Panqu",
""
],
[
"Cottrell",
"Garrison W.",
""
]
] | TITLE: Basic Level Categorization Facilitates Visual Object Recognition
ABSTRACT: Recent advances in deep learning have led to significant progress in the
computer vision field, especially for visual object recognition tasks. The
features useful for object classification are learned by feed-forward deep
convolutional neural networks (CNNs) automatically, and they are shown to be
able to predict and decode neural representations in the ventral visual pathway
of humans and monkeys. However, despite the huge amount of work on optimizing
CNNs, there has not been much research focused on linking CNNs with guiding
principles from the human visual cortex. In this work, we propose a network
optimization strategy inspired by both of the developmental trajectory of
children's visual object recognition capabilities, and Bar (2003), who
hypothesized that basic level information is carried in the fast magnocellular
pathway through the prefrontal cortex (PFC) and then projected back to inferior
temporal cortex (IT), where subordinate level categorization is achieved. We
instantiate this idea by training a deep CNN to perform basic level object
categorization first, and then train it on subordinate level categorization. We
apply this idea to training AlexNet (Krizhevsky et al., 2012) on the ILSVRC
2012 dataset and show that the top-5 accuracy increases from 80.13% to 82.14%,
demonstrating the effectiveness of the method. We also show that subsequent
transfer learning on smaller datasets gives superior results.
| no_new_dataset | 0.950869 |
1511.04306 | Sebastian Stober | Sebastian Stober, Avital Sternin, Adrian M. Owen and Jessica A. Grahn | Deep Feature Learning for EEG Recordings | submitted as conference paper for ICLR 2016 | null | null | null | cs.NE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce and compare several strategies for learning discriminative
features from electroencephalography (EEG) recordings using deep learning
techniques. EEG data are generally only available in small quantities, they are
high-dimensional with a poor signal-to-noise ratio, and there is considerable
variability between individual subjects and recording sessions. Our proposed
techniques specifically address these challenges for feature learning.
Cross-trial encoding forces auto-encoders to focus on features that are stable
across trials. Similarity-constraint encoders learn features that allow to
distinguish between classes by demanding that two trials from the same class
are more similar to each other than to trials from other classes. This
tuple-based training approach is especially suitable for small datasets.
Hydra-nets allow for separate processing pathways adapting to subsets of a
dataset and thus combine the advantages of individual feature learning (better
adaptation of early, low-level processing) with group model training (better
generalization of higher-level processing in deeper layers). This way, models
can, for instance, adapt to each subject individually to compensate for
differences in spatial patterns due to anatomical differences or variance in
electrode positions. The different techniques are evaluated using the publicly
available OpenMIIR dataset of EEG recordings taken while participants listened
to and imagined music.
| [
{
"version": "v1",
"created": "Fri, 13 Nov 2015 15:07:17 GMT"
},
{
"version": "v2",
"created": "Thu, 19 Nov 2015 22:04:12 GMT"
},
{
"version": "v3",
"created": "Fri, 27 Nov 2015 18:24:08 GMT"
},
{
"version": "v4",
"created": "Thu, 7 Jan 2016 16:26:42 GMT"
}
] | 2016-01-08T00:00:00 | [
[
"Stober",
"Sebastian",
""
],
[
"Sternin",
"Avital",
""
],
[
"Owen",
"Adrian M.",
""
],
[
"Grahn",
"Jessica A.",
""
]
] | TITLE: Deep Feature Learning for EEG Recordings
ABSTRACT: We introduce and compare several strategies for learning discriminative
features from electroencephalography (EEG) recordings using deep learning
techniques. EEG data are generally only available in small quantities, they are
high-dimensional with a poor signal-to-noise ratio, and there is considerable
variability between individual subjects and recording sessions. Our proposed
techniques specifically address these challenges for feature learning.
Cross-trial encoding forces auto-encoders to focus on features that are stable
across trials. Similarity-constraint encoders learn features that allow to
distinguish between classes by demanding that two trials from the same class
are more similar to each other than to trials from other classes. This
tuple-based training approach is especially suitable for small datasets.
Hydra-nets allow for separate processing pathways adapting to subsets of a
dataset and thus combine the advantages of individual feature learning (better
adaptation of early, low-level processing) with group model training (better
generalization of higher-level processing in deeper layers). This way, models
can, for instance, adapt to each subject individually to compensate for
differences in spatial patterns due to anatomical differences or variance in
electrode positions. The different techniques are evaluated using the publicly
available OpenMIIR dataset of EEG recordings taken while participants listened
to and imagined music.
| no_new_dataset | 0.944125 |
1601.01411 | Chetan Tonde | Chetan Tonde and Ahmed Elgammal | Learning Kernels for Structured Prediction using Polynomial Kernel
Transformations | null | null | null | 21 pages, 10 figures | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning the kernel functions used in kernel methods has been a vastly
explored area in machine learning. It is now widely accepted that to obtain
'good' performance, learning a kernel function is the key challenge. In this
work we focus on learning kernel representations for structured regression. We
propose use of polynomials expansion of kernels, referred to as Schoenberg
transforms and Gegenbaur transforms, which arise from the seminal result of
Schoenberg (1938). These kernels can be thought of as polynomial combination of
input features in a high dimensional reproducing kernel Hilbert space (RKHS).
We learn kernels over input and output for structured data, such that,
dependency between kernel features is maximized. We use Hilbert-Schmidt
Independence Criterion (HSIC) to measure this. We also give an efficient,
matrix decomposition-based algorithm to learn these kernel transformations, and
demonstrate state-of-the-art results on several real-world datasets.
| [
{
"version": "v1",
"created": "Thu, 7 Jan 2016 06:37:48 GMT"
}
] | 2016-01-08T00:00:00 | [
[
"Tonde",
"Chetan",
""
],
[
"Elgammal",
"Ahmed",
""
]
] | TITLE: Learning Kernels for Structured Prediction using Polynomial Kernel
Transformations
ABSTRACT: Learning the kernel functions used in kernel methods has been a vastly
explored area in machine learning. It is now widely accepted that to obtain
'good' performance, learning a kernel function is the key challenge. In this
work we focus on learning kernel representations for structured regression. We
propose use of polynomials expansion of kernels, referred to as Schoenberg
transforms and Gegenbaur transforms, which arise from the seminal result of
Schoenberg (1938). These kernels can be thought of as polynomial combination of
input features in a high dimensional reproducing kernel Hilbert space (RKHS).
We learn kernels over input and output for structured data, such that,
dependency between kernel features is maximized. We use Hilbert-Schmidt
Independence Criterion (HSIC) to measure this. We also give an efficient,
matrix decomposition-based algorithm to learn these kernel transformations, and
demonstrate state-of-the-art results on several real-world datasets.
| no_new_dataset | 0.947575 |
1405.3625 | Nadeem Malik A | Nadeem A. Malik | On Turbulent Particle Pair Diffusion | Submitted to J. Fluid Mechanics, 6 January, 2016. 33 pages. 9 figures | null | null | null | physics.flu-dyn math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Richardson's theory of turbulent particle pair diffusion [Richardson, L. F.
Proc. Roy. Soc. Lond. A 100, 709--737, 1926], based upon observational data, is
equivalent to a locality hypothesis in which the turbulent pair diffusivity
$(K)$ scales with the pair separation $(\sigma_l)$ with a 4/3-power law, $K\sim
\sigma_l^{4/3}$. Here, a reappraisal of the 1926 dataset reveals that one of
the data-points is from a molecular diffusion context; the remaining data from
geophysical turbulence display an unequivocal non-local scaling, $K \sim
\sigma_l^{1.564}$. Consequently, the foundations of pair diffusion theory have
been re-examined, leading to a new theory based upon the principle that both
local and non-local diffusional processes govern pair diffusion in homogeneous
turbulence. Through a novel mathematical approach the theory is developed in
the context of generalised power law energy spectra, $E(k)\sim k^{-p}$ for
$1<p\le 3$, over extended inertial subranges. The theory predicts the scaling,
$K(p)\sim \sigma_l^{\gamma_p}$, with $\gamma_p$ intermediate between the purely
local and the purely non-local scalings, i.e. $(1+p)/2<\gamma_p\le 2$. A
Lagrangian diffusion model, Kinematic Simulations [Kraichnan, R. H., Phys.
Fluids 13, 22-31, 1970; Fung et al., J. Fluid Mech. 236, 281-318, 1992], is
used to examine the predictions of the new theory all of which are confirmed.
The simulations produce the scalings, $K\sim \sigma_l^{1.545}$ to $\sim
\sigma_l^{1.570}$, in the accepted range of intermittent turbulence spectra,
$E(k)\sim k^{-1.72}$ to $\sim k^{-1.74}$, in close agreement with the revised
1926 dataset.
| [
{
"version": "v1",
"created": "Wed, 14 May 2014 19:08:14 GMT"
},
{
"version": "v2",
"created": "Thu, 31 Jul 2014 15:23:20 GMT"
},
{
"version": "v3",
"created": "Wed, 6 Jan 2016 17:54:28 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Malik",
"Nadeem A.",
""
]
] | TITLE: On Turbulent Particle Pair Diffusion
ABSTRACT: Richardson's theory of turbulent particle pair diffusion [Richardson, L. F.
Proc. Roy. Soc. Lond. A 100, 709--737, 1926], based upon observational data, is
equivalent to a locality hypothesis in which the turbulent pair diffusivity
$(K)$ scales with the pair separation $(\sigma_l)$ with a 4/3-power law, $K\sim
\sigma_l^{4/3}$. Here, a reappraisal of the 1926 dataset reveals that one of
the data-points is from a molecular diffusion context; the remaining data from
geophysical turbulence display an unequivocal non-local scaling, $K \sim
\sigma_l^{1.564}$. Consequently, the foundations of pair diffusion theory have
been re-examined, leading to a new theory based upon the principle that both
local and non-local diffusional processes govern pair diffusion in homogeneous
turbulence. Through a novel mathematical approach the theory is developed in
the context of generalised power law energy spectra, $E(k)\sim k^{-p}$ for
$1<p\le 3$, over extended inertial subranges. The theory predicts the scaling,
$K(p)\sim \sigma_l^{\gamma_p}$, with $\gamma_p$ intermediate between the purely
local and the purely non-local scalings, i.e. $(1+p)/2<\gamma_p\le 2$. A
Lagrangian diffusion model, Kinematic Simulations [Kraichnan, R. H., Phys.
Fluids 13, 22-31, 1970; Fung et al., J. Fluid Mech. 236, 281-318, 1992], is
used to examine the predictions of the new theory all of which are confirmed.
The simulations produce the scalings, $K\sim \sigma_l^{1.545}$ to $\sim
\sigma_l^{1.570}$, in the accepted range of intermittent turbulence spectra,
$E(k)\sim k^{-1.72}$ to $\sim k^{-1.74}$, in close agreement with the revised
1926 dataset.
| no_new_dataset | 0.949716 |
1408.3264 | Mohammad Ali Keyvanrad | Mohammad Ali Keyvanrad, Mohammad Mehdi Homayounpour | A brief survey on deep belief networks and introducing a new object
oriented toolbox (DeeBNet) | Technical Report 27 pages, Ver3.0 | null | null | null | cs.CV cs.LG cs.MS cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nowadays, this is very popular to use the deep architectures in machine
learning. Deep Belief Networks (DBNs) are deep architectures that use stack of
Restricted Boltzmann Machines (RBM) to create a powerful generative model using
training data. DBNs have many ability like feature extraction and
classification that are used in many applications like image processing, speech
processing and etc. This paper introduces a new object oriented MATLAB toolbox
with most of abilities needed for the implementation of DBNs. In the new
version, the toolbox can be used in Octave. According to the results of the
experiments conducted on MNIST (image), ISOLET (speech), and 20 Newsgroups
(text) datasets, it was shown that the toolbox can learn automatically a good
representation of the input from unlabeled data with better discrimination
between different classes. Also on all datasets, the obtained classification
errors are comparable to those of state of the art classifiers. In addition,
the toolbox supports different sampling methods (e.g. Gibbs, CD, PCD and our
new FEPCD method), different sparsity methods (quadratic, rate distortion and
our new normal method), different RBM types (generative and discriminative),
using GPU, etc. The toolbox is a user-friendly open source software and is
freely available on the website
http://ceit.aut.ac.ir/~keyvanrad/DeeBNet%20Toolbox.html .
| [
{
"version": "v1",
"created": "Thu, 14 Aug 2014 12:37:57 GMT"
},
{
"version": "v2",
"created": "Mon, 8 Dec 2014 14:44:02 GMT"
},
{
"version": "v3",
"created": "Thu, 9 Jul 2015 12:44:01 GMT"
},
{
"version": "v4",
"created": "Fri, 10 Jul 2015 13:21:02 GMT"
},
{
"version": "v5",
"created": "Wed, 22 Jul 2015 14:25:13 GMT"
},
{
"version": "v6",
"created": "Mon, 7 Sep 2015 14:44:47 GMT"
},
{
"version": "v7",
"created": "Wed, 6 Jan 2016 13:20:11 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Keyvanrad",
"Mohammad Ali",
""
],
[
"Homayounpour",
"Mohammad Mehdi",
""
]
] | TITLE: A brief survey on deep belief networks and introducing a new object
oriented toolbox (DeeBNet)
ABSTRACT: Nowadays, this is very popular to use the deep architectures in machine
learning. Deep Belief Networks (DBNs) are deep architectures that use stack of
Restricted Boltzmann Machines (RBM) to create a powerful generative model using
training data. DBNs have many ability like feature extraction and
classification that are used in many applications like image processing, speech
processing and etc. This paper introduces a new object oriented MATLAB toolbox
with most of abilities needed for the implementation of DBNs. In the new
version, the toolbox can be used in Octave. According to the results of the
experiments conducted on MNIST (image), ISOLET (speech), and 20 Newsgroups
(text) datasets, it was shown that the toolbox can learn automatically a good
representation of the input from unlabeled data with better discrimination
between different classes. Also on all datasets, the obtained classification
errors are comparable to those of state of the art classifiers. In addition,
the toolbox supports different sampling methods (e.g. Gibbs, CD, PCD and our
new FEPCD method), different sparsity methods (quadratic, rate distortion and
our new normal method), different RBM types (generative and discriminative),
using GPU, etc. The toolbox is a user-friendly open source software and is
freely available on the website
http://ceit.aut.ac.ir/~keyvanrad/DeeBNet%20Toolbox.html .
| no_new_dataset | 0.94366 |
1506.01497 | Kaiming He | Shaoqing Ren, Kaiming He, Ross Girshick, Jian Sun | Faster R-CNN: Towards Real-Time Object Detection with Region Proposal
Networks | Extended tech report | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | State-of-the-art object detection networks depend on region proposal
algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN
have reduced the running time of these detection networks, exposing region
proposal computation as a bottleneck. In this work, we introduce a Region
Proposal Network (RPN) that shares full-image convolutional features with the
detection network, thus enabling nearly cost-free region proposals. An RPN is a
fully convolutional network that simultaneously predicts object bounds and
objectness scores at each position. The RPN is trained end-to-end to generate
high-quality region proposals, which are used by Fast R-CNN for detection. We
further merge RPN and Fast R-CNN into a single network by sharing their
convolutional features---using the recently popular terminology of neural
networks with 'attention' mechanisms, the RPN component tells the unified
network where to look. For the very deep VGG-16 model, our detection system has
a frame rate of 5fps (including all steps) on a GPU, while achieving
state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS
COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015
competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning
entries in several tracks. Code has been made publicly available.
| [
{
"version": "v1",
"created": "Thu, 4 Jun 2015 07:58:34 GMT"
},
{
"version": "v2",
"created": "Sun, 13 Sep 2015 07:54:00 GMT"
},
{
"version": "v3",
"created": "Wed, 6 Jan 2016 06:30:17 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Ren",
"Shaoqing",
""
],
[
"He",
"Kaiming",
""
],
[
"Girshick",
"Ross",
""
],
[
"Sun",
"Jian",
""
]
] | TITLE: Faster R-CNN: Towards Real-Time Object Detection with Region Proposal
Networks
ABSTRACT: State-of-the-art object detection networks depend on region proposal
algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN
have reduced the running time of these detection networks, exposing region
proposal computation as a bottleneck. In this work, we introduce a Region
Proposal Network (RPN) that shares full-image convolutional features with the
detection network, thus enabling nearly cost-free region proposals. An RPN is a
fully convolutional network that simultaneously predicts object bounds and
objectness scores at each position. The RPN is trained end-to-end to generate
high-quality region proposals, which are used by Fast R-CNN for detection. We
further merge RPN and Fast R-CNN into a single network by sharing their
convolutional features---using the recently popular terminology of neural
networks with 'attention' mechanisms, the RPN component tells the unified
network where to look. For the very deep VGG-16 model, our detection system has
a frame rate of 5fps (including all steps) on a GPU, while achieving
state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS
COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015
competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning
entries in several tracks. Code has been made publicly available.
| no_new_dataset | 0.951863 |
1507.07595 | Tengyu Ma | Jason D. Lee, Qihang Lin, Tengyu Ma, Tianbao Yang | Distributed Stochastic Variance Reduced Gradient Methods and A Lower
Bound for Communication Complexity | significant addition to both theory and experimental results | null | null | null | math.OC cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study distributed optimization algorithms for minimizing the average of
convex functions. The applications include empirical risk minimization problems
in statistical machine learning where the datasets are large and have to be
stored on different machines. We design a distributed stochastic variance
reduced gradient algorithm that, under certain conditions on the condition
number, simultaneously achieves the optimal parallel runtime, amount of
communication and rounds of communication among all distributed first-order
methods up to constant factors. Our method and its accelerated extension also
outperform existing distributed algorithms in terms of the rounds of
communication as long as the condition number is not too large compared to the
size of data in each machine. We also prove a lower bound for the number of
rounds of communication for a broad class of distributed first-order methods
including the proposed algorithms in this paper. We show that our accelerated
distributed stochastic variance reduced gradient algorithm achieves this lower
bound so that it uses the fewest rounds of communication among all distributed
first-order algorithms.
| [
{
"version": "v1",
"created": "Mon, 27 Jul 2015 22:09:57 GMT"
},
{
"version": "v2",
"created": "Wed, 6 Jan 2016 19:26:31 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Lee",
"Jason D.",
""
],
[
"Lin",
"Qihang",
""
],
[
"Ma",
"Tengyu",
""
],
[
"Yang",
"Tianbao",
""
]
] | TITLE: Distributed Stochastic Variance Reduced Gradient Methods and A Lower
Bound for Communication Complexity
ABSTRACT: We study distributed optimization algorithms for minimizing the average of
convex functions. The applications include empirical risk minimization problems
in statistical machine learning where the datasets are large and have to be
stored on different machines. We design a distributed stochastic variance
reduced gradient algorithm that, under certain conditions on the condition
number, simultaneously achieves the optimal parallel runtime, amount of
communication and rounds of communication among all distributed first-order
methods up to constant factors. Our method and its accelerated extension also
outperform existing distributed algorithms in terms of the rounds of
communication as long as the condition number is not too large compared to the
size of data in each machine. We also prove a lower bound for the number of
rounds of communication for a broad class of distributed first-order methods
including the proposed algorithms in this paper. We show that our accelerated
distributed stochastic variance reduced gradient algorithm achieves this lower
bound so that it uses the fewest rounds of communication among all distributed
first-order algorithms.
| no_new_dataset | 0.944228 |
1511.02490 | Chris Cummins | Chris Cummins, Pavlos Petoumenos, Michel Steuwer, and Hugh Leather | Autotuning OpenCL Workgroup Size for Stencil Patterns | 8 pages, 6 figures, presented at the 6th International Workshop on
Adaptive Self-tuning Computing Systems (ADAPT '16) | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Selecting an appropriate workgroup size is critical for the performance of
OpenCL kernels, and requires knowledge of the underlying hardware, the data
being operated on, and the implementation of the kernel. This makes portable
performance of OpenCL programs a challenging goal, since simple heuristics and
statically chosen values fail to exploit the available performance. To address
this, we propose the use of machine learning-enabled autotuning to
automatically predict workgroup sizes for stencil patterns on CPUs and
multi-GPUs.
We present three methodologies for predicting workgroup sizes. The first,
using classifiers to select the optimal workgroup size. The second and third
proposed methodologies employ the novel use of regressors for performing
classification by predicting the runtime of kernels and the relative
performance of different workgroup sizes, respectively. We evaluate the
effectiveness of each technique in an empirical study of 429 combinations of
architecture, kernel, and dataset, comparing an average of 629 different
workgroup sizes for each. We find that autotuning provides a median 3.79x
speedup over the best possible fixed workgroup size, achieving 94% of the
maximum performance.
| [
{
"version": "v1",
"created": "Sun, 8 Nov 2015 14:56:12 GMT"
},
{
"version": "v2",
"created": "Sun, 22 Nov 2015 23:22:04 GMT"
},
{
"version": "v3",
"created": "Wed, 6 Jan 2016 15:50:33 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Cummins",
"Chris",
""
],
[
"Petoumenos",
"Pavlos",
""
],
[
"Steuwer",
"Michel",
""
],
[
"Leather",
"Hugh",
""
]
] | TITLE: Autotuning OpenCL Workgroup Size for Stencil Patterns
ABSTRACT: Selecting an appropriate workgroup size is critical for the performance of
OpenCL kernels, and requires knowledge of the underlying hardware, the data
being operated on, and the implementation of the kernel. This makes portable
performance of OpenCL programs a challenging goal, since simple heuristics and
statically chosen values fail to exploit the available performance. To address
this, we propose the use of machine learning-enabled autotuning to
automatically predict workgroup sizes for stencil patterns on CPUs and
multi-GPUs.
We present three methodologies for predicting workgroup sizes. The first,
using classifiers to select the optimal workgroup size. The second and third
proposed methodologies employ the novel use of regressors for performing
classification by predicting the runtime of kernels and the relative
performance of different workgroup sizes, respectively. We evaluate the
effectiveness of each technique in an empirical study of 429 combinations of
architecture, kernel, and dataset, comparing an average of 629 different
workgroup sizes for each. We find that autotuning provides a median 3.79x
speedup over the best possible fixed workgroup size, achieving 94% of the
maximum performance.
| no_new_dataset | 0.949201 |
1601.00978 | Joseph Paul Cohen | Joseph Paul Cohen and Henry Z. Lo and Tingting Lu and Wei Ding | Crater Detection via Convolutional Neural Networks | 2 Pages. Submitted to 47th Lunar and Planetary Science Conference
(LPSC 2016) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Craters are among the most studied geomorphic features in the Solar System
because they yield important information about the past and present geological
processes and provide information about the relative ages of observed geologic
formations. We present a method for automatic crater detection using advanced
machine learning to deal with the large amount of satellite imagery collected.
The challenge of automatically detecting craters comes from their is complex
surface because their shape erodes over time to blend into the surface.
Bandeira provided a seminal dataset that embodied this challenge that is still
an unsolved pattern recognition problem to this day. There has been work to
solve this challenge based on extracting shape and contrast features and then
applying classification models on those features. The limiting factor in this
existing work is the use of hand crafted filters on the image such as Gabor or
Sobel filters or Haar features. These hand crafted methods rely on domain
knowledge to construct. We would like to learn the optimal filters and features
based on training examples. In order to dynamically learn filters and features
we look to Convolutional Neural Networks (CNNs) which have shown their
dominance in computer vision. The power of CNNs is that they can learn image
filters which generate features for high accuracy classification.
| [
{
"version": "v1",
"created": "Tue, 5 Jan 2016 21:03:59 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Cohen",
"Joseph Paul",
""
],
[
"Lo",
"Henry Z.",
""
],
[
"Lu",
"Tingting",
""
],
[
"Ding",
"Wei",
""
]
] | TITLE: Crater Detection via Convolutional Neural Networks
ABSTRACT: Craters are among the most studied geomorphic features in the Solar System
because they yield important information about the past and present geological
processes and provide information about the relative ages of observed geologic
formations. We present a method for automatic crater detection using advanced
machine learning to deal with the large amount of satellite imagery collected.
The challenge of automatically detecting craters comes from their is complex
surface because their shape erodes over time to blend into the surface.
Bandeira provided a seminal dataset that embodied this challenge that is still
an unsolved pattern recognition problem to this day. There has been work to
solve this challenge based on extracting shape and contrast features and then
applying classification models on those features. The limiting factor in this
existing work is the use of hand crafted filters on the image such as Gabor or
Sobel filters or Haar features. These hand crafted methods rely on domain
knowledge to construct. We would like to learn the optimal filters and features
based on training examples. In order to dynamically learn filters and features
we look to Convolutional Neural Networks (CNNs) which have shown their
dominance in computer vision. The power of CNNs is that they can learn image
filters which generate features for high accuracy classification.
| new_dataset | 0.743075 |
1601.00998 | Alexandre Robicquet Alexandre Robicquet | Alexandre Robicquet, Alexandre Alahi, Amir Sadeghian, Bryan Anenberg,
John Doherty, Eli Wu, and Silvio Savarese | Forecasting Social Navigation in Crowded Complex Scenes | null | null | null | null | cs.CV cs.RO cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When humans navigate a crowed space such as a university campus or the
sidewalks of a busy street, they follow common sense rules based on social
etiquette. In this paper, we argue that in order to enable the design of new
algorithms that can take fully advantage of these rules to better solve tasks
such as target tracking or trajectory forecasting, we need to have access to
better data in the first place. To that end, we contribute the very first large
scale dataset (to the best of our knowledge) that collects images and videos of
various types of targets (not just pedestrians, but also bikers, skateboarders,
cars, buses, golf carts) that navigate in a real-world outdoor environment such
as a university campus. We present an extensive evaluation where different
methods for trajectory forecasting are evaluated and compared. Moreover, we
present a new algorithm for trajectory prediction that exploits the complexity
of our new dataset and allows to: i) incorporate inter-class interactions into
trajectory prediction models (e.g, pedestrian vs bike) as opposed to just
intra-class interactions (e.g., pedestrian vs pedestrian); ii) model the degree
to which the social forces are regulating an interaction. We call the latter
"social sensitivity"and it captures the sensitivity to which a target is
responding to a certain interaction. An extensive experimental evaluation
demonstrates the effectiveness of our novel approach.
| [
{
"version": "v1",
"created": "Tue, 5 Jan 2016 22:10:15 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Robicquet",
"Alexandre",
""
],
[
"Alahi",
"Alexandre",
""
],
[
"Sadeghian",
"Amir",
""
],
[
"Anenberg",
"Bryan",
""
],
[
"Doherty",
"John",
""
],
[
"Wu",
"Eli",
""
],
[
"Savarese",
"Silvio",
""
]
] | TITLE: Forecasting Social Navigation in Crowded Complex Scenes
ABSTRACT: When humans navigate a crowed space such as a university campus or the
sidewalks of a busy street, they follow common sense rules based on social
etiquette. In this paper, we argue that in order to enable the design of new
algorithms that can take fully advantage of these rules to better solve tasks
such as target tracking or trajectory forecasting, we need to have access to
better data in the first place. To that end, we contribute the very first large
scale dataset (to the best of our knowledge) that collects images and videos of
various types of targets (not just pedestrians, but also bikers, skateboarders,
cars, buses, golf carts) that navigate in a real-world outdoor environment such
as a university campus. We present an extensive evaluation where different
methods for trajectory forecasting are evaluated and compared. Moreover, we
present a new algorithm for trajectory prediction that exploits the complexity
of our new dataset and allows to: i) incorporate inter-class interactions into
trajectory prediction models (e.g, pedestrian vs bike) as opposed to just
intra-class interactions (e.g., pedestrian vs pedestrian); ii) model the degree
to which the social forces are regulating an interaction. We call the latter
"social sensitivity"and it captures the sensitivity to which a target is
responding to a certain interaction. An extensive experimental evaluation
demonstrates the effectiveness of our novel approach.
| new_dataset | 0.964589 |
1601.01100 | Guo Qiang | Guo Qiang, Tu Dan, Li Guohui, Lei Jun | Memory Matters: Convolutional Recurrent Neural Network for Scene Text
Recognition | 6 pages, 2 figures, 2 tables | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Text recognition in natural scene is a challenging problem due to the many
factors affecting text appearance. In this paper, we presents a method that
directly transcribes scene text images to text without needing of sophisticated
character segmentation. We leverage recent advances of deep neural networks to
model the appearance of scene text images with temporal dynamics. Specifically,
we integrates convolutional neural network (CNN) and recurrent neural network
(RNN) which is motivated by observing the complementary modeling capabilities
of the two models. The main contribution of this work is investigating how
temporal memory helps in an segmentation free fashion for this specific
problem. By using long short-term memory (LSTM) blocks as hidden units, our
model can retain long-term memory compared with HMMs which only maintain
short-term state dependences. We conduct experiments on Street View House
Number dataset containing highly variable number images. The results
demonstrate the superiority of the proposed method over traditional HMM based
methods.
| [
{
"version": "v1",
"created": "Wed, 6 Jan 2016 07:36:15 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Qiang",
"Guo",
""
],
[
"Dan",
"Tu",
""
],
[
"Guohui",
"Li",
""
],
[
"Jun",
"Lei",
""
]
] | TITLE: Memory Matters: Convolutional Recurrent Neural Network for Scene Text
Recognition
ABSTRACT: Text recognition in natural scene is a challenging problem due to the many
factors affecting text appearance. In this paper, we presents a method that
directly transcribes scene text images to text without needing of sophisticated
character segmentation. We leverage recent advances of deep neural networks to
model the appearance of scene text images with temporal dynamics. Specifically,
we integrates convolutional neural network (CNN) and recurrent neural network
(RNN) which is motivated by observing the complementary modeling capabilities
of the two models. The main contribution of this work is investigating how
temporal memory helps in an segmentation free fashion for this specific
problem. By using long short-term memory (LSTM) blocks as hidden units, our
model can retain long-term memory compared with HMMs which only maintain
short-term state dependences. We conduct experiments on Street View House
Number dataset containing highly variable number images. The results
demonstrate the superiority of the proposed method over traditional HMM based
methods.
| no_new_dataset | 0.950227 |
1601.01191 | Fabien Mathieu | The Dang Huynh (LINCS), Fabien Mathieu (LINCS), Laurent Viennot (GANG,
LINCS) | LiveRank: How to Refresh Old Datasets | null | null | 10.1080/15427951.2015.1098756 | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper considers the problem of refreshing a dataset. More precisely ,
given a collection of nodes gathered at some time (Web pages, users from an
online social network) along with some structure (hyperlinks, social
relationships), we want to identify a significant fraction of the nodes that
still exist at present time. The liveness of an old node can be tested through
an online query at present time. We call LiveRank a ranking of the old pages so
that active nodes are more likely to appear first. The quality of a LiveRank is
measured by the number of queries necessary to identify a given fraction of the
active nodes when using the LiveRank order. We study different scenarios from a
static setting where the Liv-eRank is computed before any query is made, to
dynamic settings where the LiveRank can be updated as queries are processed.
Our results show that building on the PageRank can lead to efficient LiveRanks,
for Web graphs as well as for online social networks.
| [
{
"version": "v1",
"created": "Wed, 6 Jan 2016 14:25:23 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Huynh",
"The Dang",
"",
"LINCS"
],
[
"Mathieu",
"Fabien",
"",
"LINCS"
],
[
"Viennot",
"Laurent",
"",
"GANG,\n LINCS"
]
] | TITLE: LiveRank: How to Refresh Old Datasets
ABSTRACT: This paper considers the problem of refreshing a dataset. More precisely ,
given a collection of nodes gathered at some time (Web pages, users from an
online social network) along with some structure (hyperlinks, social
relationships), we want to identify a significant fraction of the nodes that
still exist at present time. The liveness of an old node can be tested through
an online query at present time. We call LiveRank a ranking of the old pages so
that active nodes are more likely to appear first. The quality of a LiveRank is
measured by the number of queries necessary to identify a given fraction of the
active nodes when using the LiveRank order. We study different scenarios from a
static setting where the Liv-eRank is computed before any query is made, to
dynamic settings where the LiveRank can be updated as queries are processed.
Our results show that building on the PageRank can lead to efficient LiveRanks,
for Web graphs as well as for online social networks.
| no_new_dataset | 0.942665 |
1601.01195 | Kamal Sarkar | Kamal Sarkar | Part-of-Speech Tagging for Code-mixed Indian Social Media Text at ICON
2015 | NLP Tool Contest on "POS Tagging For Code-mixed Indian Social Media
Text" held in conjunction with International Conference on Natural Language
Processing(ICON 2015). arXiv admin note: text overlap with arXiv:1512.03950,
arXiv:1405.7397 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper discusses the experiments carried out by us at Jadavpur University
as part of the participation in ICON 2015 task: POS Tagging for Code-mixed
Indian Social Media Text. The tool that we have developed for the task is based
on Trigram Hidden Markov Model that utilizes information from dictionary as
well as some other word level features to enhance the observation probabilities
of the known tokens as well as unknown tokens. We submitted runs for
Bengali-English, Hindi-English and Tamil-English Language pairs. Our system has
been trained and tested on the datasets released for ICON 2015 shared task: POS
Tagging For Code-mixed Indian Social Media Text. In constrained mode, our
system obtains average overall accuracy (averaged over all three language
pairs) of 75.60% which is very close to other participating two systems (76.79%
for IIITH and 75.79% for AMRITA_CEN) ranked higher than our system. In
unconstrained mode, our system obtains average overall accuracy of 70.65% which
is also close to the system (72.85% for AMRITA_CEN) which obtains the highest
average overall accuracy.
| [
{
"version": "v1",
"created": "Wed, 6 Jan 2016 14:40:38 GMT"
}
] | 2016-01-07T00:00:00 | [
[
"Sarkar",
"Kamal",
""
]
] | TITLE: Part-of-Speech Tagging for Code-mixed Indian Social Media Text at ICON
2015
ABSTRACT: This paper discusses the experiments carried out by us at Jadavpur University
as part of the participation in ICON 2015 task: POS Tagging for Code-mixed
Indian Social Media Text. The tool that we have developed for the task is based
on Trigram Hidden Markov Model that utilizes information from dictionary as
well as some other word level features to enhance the observation probabilities
of the known tokens as well as unknown tokens. We submitted runs for
Bengali-English, Hindi-English and Tamil-English Language pairs. Our system has
been trained and tested on the datasets released for ICON 2015 shared task: POS
Tagging For Code-mixed Indian Social Media Text. In constrained mode, our
system obtains average overall accuracy (averaged over all three language
pairs) of 75.60% which is very close to other participating two systems (76.79%
for IIITH and 75.79% for AMRITA_CEN) ranked higher than our system. In
unconstrained mode, our system obtains average overall accuracy of 70.65% which
is also close to the system (72.85% for AMRITA_CEN) which obtains the highest
average overall accuracy.
| no_new_dataset | 0.957557 |
1309.1785 | Eduardo Graells-Garrido | Eduardo Graells-Garrido and Barbara Poblete | #Santiago is not #Chile, or is it? A Model to Normalize Social Media
Impact | Accepted in ChileCHI 2013, I Chilean Conference on Human-Computer
Interaction | null | 10.1145/2535597.2535611 | null | cs.SI physics.soc-ph | http://creativecommons.org/licenses/by-nc-sa/3.0/ | Online social networks are known to be demographically biased. Currently
there are questions about what degree of representativity of the physical
population they have, and how population biases impact user-generated content.
In this paper we focus on centralism, a problem affecting Chile. Assuming that
local differences exist in a country, in terms of vocabulary, we built a
methodology based on the vector space model to find distinctive content from
different locations, and use it to create classifiers to predict whether the
content of a micro-post is related to a particular location, having in mind a
geographically diverse selection of micro-posts. We evaluate them in a case
study where we analyze the virtual population of Chile that participated in the
Twitter social network during an event of national relevance: the municipal
(local governments) elections held in 2012. We observe that the participating
virtual population is spatially representative of the physical population,
implying that there is centralism in Twitter. Our classifiers out-perform a non
geographically-diverse baseline at the regional level, and have the same
accuracy at a provincial level. However, our approach makes assumptions that
need to be tested in multi-thematic and more general datasets. We leave this
for future work.
| [
{
"version": "v1",
"created": "Fri, 6 Sep 2013 21:58:30 GMT"
}
] | 2016-01-06T00:00:00 | [
[
"Graells-Garrido",
"Eduardo",
""
],
[
"Poblete",
"Barbara",
""
]
] | TITLE: #Santiago is not #Chile, or is it? A Model to Normalize Social Media
Impact
ABSTRACT: Online social networks are known to be demographically biased. Currently
there are questions about what degree of representativity of the physical
population they have, and how population biases impact user-generated content.
In this paper we focus on centralism, a problem affecting Chile. Assuming that
local differences exist in a country, in terms of vocabulary, we built a
methodology based on the vector space model to find distinctive content from
different locations, and use it to create classifiers to predict whether the
content of a micro-post is related to a particular location, having in mind a
geographically diverse selection of micro-posts. We evaluate them in a case
study where we analyze the virtual population of Chile that participated in the
Twitter social network during an event of national relevance: the municipal
(local governments) elections held in 2012. We observe that the participating
virtual population is spatially representative of the physical population,
implying that there is centralism in Twitter. Our classifiers out-perform a non
geographically-diverse baseline at the regional level, and have the same
accuracy at a provincial level. However, our approach makes assumptions that
need to be tested in multi-thematic and more general datasets. We leave this
for future work.
| no_new_dataset | 0.942348 |
1502.07310 | Amy Yu | Amy Zhao Yu, Shahar Ronen, Kevin Hu, Tiffany Lu, C\'esar A. Hidalgo | Pantheon 1.0, a manually verified dataset of globally famous biographies | Scientific Data 2:150075 | null | 10.1038/sdata.2015.75 | null | physics.soc-ph cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present the Pantheon 1.0 dataset: a manually verified dataset of
individuals that have transcended linguistic, temporal, and geographic
boundaries. The Pantheon 1.0 dataset includes the 11,341 biographies present in
more than 25 languages in Wikipedia and is enriched with: (i) manually verified
demographic information (place and date of birth, gender) (ii) a taxonomy of
occupations classifying each biography at three levels of aggregation and (iii)
two measures of global popularity including the number of languages in which a
biography is present in Wikipedia (L), and the Historical Popularity Index
(HPI) a metric that combines information on L, time since birth, and page-views
(2008-2013). We compare the Pantheon 1.0 dataset to data from the 2003 book,
Human Accomplishments, and also to external measures of accomplishment in
individual games and sports: Tennis, Swimming, Car Racing, and Chess. In all of
these cases we find that measures of popularity (L and HPI) correlate highly
with individual accomplishment, suggesting that measures of global popularity
proxy the historical impact of individuals.
| [
{
"version": "v1",
"created": "Wed, 25 Feb 2015 19:17:14 GMT"
},
{
"version": "v2",
"created": "Tue, 5 Jan 2016 15:08:28 GMT"
}
] | 2016-01-06T00:00:00 | [
[
"Yu",
"Amy Zhao",
""
],
[
"Ronen",
"Shahar",
""
],
[
"Hu",
"Kevin",
""
],
[
"Lu",
"Tiffany",
""
],
[
"Hidalgo",
"César A.",
""
]
] | TITLE: Pantheon 1.0, a manually verified dataset of globally famous biographies
ABSTRACT: We present the Pantheon 1.0 dataset: a manually verified dataset of
individuals that have transcended linguistic, temporal, and geographic
boundaries. The Pantheon 1.0 dataset includes the 11,341 biographies present in
more than 25 languages in Wikipedia and is enriched with: (i) manually verified
demographic information (place and date of birth, gender) (ii) a taxonomy of
occupations classifying each biography at three levels of aggregation and (iii)
two measures of global popularity including the number of languages in which a
biography is present in Wikipedia (L), and the Historical Popularity Index
(HPI) a metric that combines information on L, time since birth, and page-views
(2008-2013). We compare the Pantheon 1.0 dataset to data from the 2003 book,
Human Accomplishments, and also to external measures of accomplishment in
individual games and sports: Tennis, Swimming, Car Racing, and Chess. In all of
these cases we find that measures of popularity (L and HPI) correlate highly
with individual accomplishment, suggesting that measures of global popularity
proxy the historical impact of individuals.
| new_dataset | 0.970465 |
1506.06221 | Pritheega Magalingam | Pritheega Magalingam, Stephen Davis, Asha Rao | Ranking the Importance Level of Intermediaries to a Criminal using a
Reliance Measure | Paper version 3.0 | null | null | null | cs.SI physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent research on finding important intermediate nodes in a network
suspected to contain criminal activity is highly dependent on network
centrality values. Betweenness centrality, for example, is widely used to rank
the nodes that act as brokers in the shortest paths connecting all source and
all the end nodes in a network. However both the shortest path node betweenness
and the linearly scaled betweenness can only show rankings for all the nodes in
a network. In this paper we explore the mathematical concept of pair-dependency
on intermediate nodes, adapting the concept to criminal relationships and
introducing a new source-intermediate reliance measure. To illustrate our
measure, we apply it to rank the nodes in the Enron email dataset and the
Noordin Top Terrorist networks. We compare the reliance ranking with Google
PageRank, Markov centrality as well as betweenness centrality and show that a
criminal investigation using the reliance measure, will lead to a different
prioritisation in terms of possible people to investigate. While the ranking
for the Noordin Top terrorist network nodes yields more extreme differences
than for the Enron email transaction network, in the latter the reliance values
for the set of finance managers immediately identified another employee
convicted of money laundering.
| [
{
"version": "v1",
"created": "Sat, 20 Jun 2015 10:04:57 GMT"
},
{
"version": "v2",
"created": "Tue, 7 Jul 2015 08:50:19 GMT"
},
{
"version": "v3",
"created": "Tue, 5 Jan 2016 02:36:17 GMT"
}
] | 2016-01-06T00:00:00 | [
[
"Magalingam",
"Pritheega",
""
],
[
"Davis",
"Stephen",
""
],
[
"Rao",
"Asha",
""
]
] | TITLE: Ranking the Importance Level of Intermediaries to a Criminal using a
Reliance Measure
ABSTRACT: Recent research on finding important intermediate nodes in a network
suspected to contain criminal activity is highly dependent on network
centrality values. Betweenness centrality, for example, is widely used to rank
the nodes that act as brokers in the shortest paths connecting all source and
all the end nodes in a network. However both the shortest path node betweenness
and the linearly scaled betweenness can only show rankings for all the nodes in
a network. In this paper we explore the mathematical concept of pair-dependency
on intermediate nodes, adapting the concept to criminal relationships and
introducing a new source-intermediate reliance measure. To illustrate our
measure, we apply it to rank the nodes in the Enron email dataset and the
Noordin Top Terrorist networks. We compare the reliance ranking with Google
PageRank, Markov centrality as well as betweenness centrality and show that a
criminal investigation using the reliance measure, will lead to a different
prioritisation in terms of possible people to investigate. While the ranking
for the Noordin Top terrorist network nodes yields more extreme differences
than for the Enron email transaction network, in the latter the reliance values
for the set of finance managers immediately identified another employee
convicted of money laundering.
| no_new_dataset | 0.953665 |
1601.00022 | Hongzhi Li | Hongzhi Li, Joseph G. Ellis, Shih-Fu Chang | Event Specific Multimodal Pattern Mining with Image-Caption Pairs | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we describe a novel framework and algorithms for discovering
image patch patterns from a large corpus of weakly supervised image-caption
pairs generated from news events. Current pattern mining techniques attempt to
find patterns that are representative and discriminative, we stipulate that our
discovered patterns must also be recognizable by humans and preferably with
meaningful names. We propose a new multimodal pattern mining approach that
leverages the descriptive captions often accompanying news images to learn
semantically meaningful image patch patterns. The mutltimodal patterns are then
named using words mined from the associated image captions for each pattern. A
novel evaluation framework is provided that demonstrates our patterns are 26.2%
more semantically meaningful than those discovered by the state of the art
vision only pipeline, and that we can provide tags for the discovered images
patches with 54.5% accuracy with no direct supervision. Our methods also
discover named patterns beyond those covered by the existing image datasets
like ImageNet. To the best of our knowledge this is the first algorithm
developed to automatically mine image patch patterns that have strong semantic
meaning specific to high-level news events, and then evaluate these patterns
based on that criteria.
| [
{
"version": "v1",
"created": "Thu, 31 Dec 2015 22:14:22 GMT"
},
{
"version": "v2",
"created": "Tue, 5 Jan 2016 01:55:22 GMT"
}
] | 2016-01-06T00:00:00 | [
[
"Li",
"Hongzhi",
""
],
[
"Ellis",
"Joseph G.",
""
],
[
"Chang",
"Shih-Fu",
""
]
] | TITLE: Event Specific Multimodal Pattern Mining with Image-Caption Pairs
ABSTRACT: In this paper we describe a novel framework and algorithms for discovering
image patch patterns from a large corpus of weakly supervised image-caption
pairs generated from news events. Current pattern mining techniques attempt to
find patterns that are representative and discriminative, we stipulate that our
discovered patterns must also be recognizable by humans and preferably with
meaningful names. We propose a new multimodal pattern mining approach that
leverages the descriptive captions often accompanying news images to learn
semantically meaningful image patch patterns. The mutltimodal patterns are then
named using words mined from the associated image captions for each pattern. A
novel evaluation framework is provided that demonstrates our patterns are 26.2%
more semantically meaningful than those discovered by the state of the art
vision only pipeline, and that we can provide tags for the discovered images
patches with 54.5% accuracy with no direct supervision. Our methods also
discover named patterns beyond those covered by the existing image datasets
like ImageNet. To the best of our knowledge this is the first algorithm
developed to automatically mine image patch patterns that have strong semantic
meaning specific to high-level news events, and then evaluate these patterns
based on that criteria.
| no_new_dataset | 0.95388 |
1601.00706 | Jimei Yang | Jimei Yang, Scott Reed, Ming-Hsuan Yang, Honglak Lee | Weakly-supervised Disentangling with Recurrent Transformations for 3D
View Synthesis | This was published in NIPS 2015 conference | null | null | null | cs.LG cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | An important problem for both graphics and vision is to synthesize novel
views of a 3D object from a single image. This is particularly challenging due
to the partial observability inherent in projecting a 3D object onto the image
space, and the ill-posedness of inferring object shape and pose. However, we
can train a neural network to address the problem if we restrict our attention
to specific object categories (in our case faces and chairs) for which we can
gather ample training data. In this paper, we propose a novel recurrent
convolutional encoder-decoder network that is trained end-to-end on the task of
rendering rotated objects starting from a single image. The recurrent structure
allows our model to capture long-term dependencies along a sequence of
transformations. We demonstrate the quality of its predictions for human faces
on the Multi-PIE dataset and for a dataset of 3D chair models, and also show
its ability to disentangle latent factors of variation (e.g., identity and
pose) without using full supervision.
| [
{
"version": "v1",
"created": "Tue, 5 Jan 2016 00:08:09 GMT"
}
] | 2016-01-06T00:00:00 | [
[
"Yang",
"Jimei",
""
],
[
"Reed",
"Scott",
""
],
[
"Yang",
"Ming-Hsuan",
""
],
[
"Lee",
"Honglak",
""
]
] | TITLE: Weakly-supervised Disentangling with Recurrent Transformations for 3D
View Synthesis
ABSTRACT: An important problem for both graphics and vision is to synthesize novel
views of a 3D object from a single image. This is particularly challenging due
to the partial observability inherent in projecting a 3D object onto the image
space, and the ill-posedness of inferring object shape and pose. However, we
can train a neural network to address the problem if we restrict our attention
to specific object categories (in our case faces and chairs) for which we can
gather ample training data. In this paper, we propose a novel recurrent
convolutional encoder-decoder network that is trained end-to-end on the task of
rendering rotated objects starting from a single image. The recurrent structure
allows our model to capture long-term dependencies along a sequence of
transformations. We demonstrate the quality of its predictions for human faces
on the Multi-PIE dataset and for a dataset of 3D chair models, and also show
its ability to disentangle latent factors of variation (e.g., identity and
pose) without using full supervision.
| no_new_dataset | 0.942929 |
1601.00825 | Concetto Spampinato Dr | Simone Palazzo, Concetto Spampinato and Daniela Giordano | Gamifying Video Object Segmentation | Submitted to PAMI | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Video object segmentation can be considered as one of the most challenging
computer vision problems. Indeed, so far, no existing solution is able to
effectively deal with the peculiarities of real-world videos, especially in
cases of articulated motion and object occlusions; limitations that appear more
evident when we compare their performance with the human one. However, manually
segmenting objects in videos is largely impractical as it requires a lot of
human time and concentration. To address this problem, in this paper we propose
an interactive video object segmentation method, which exploits, on one hand,
the capability of humans to identify correctly objects in visual scenes, and on
the other hand, the collective human brainpower to solve challenging tasks. In
particular, our method relies on a web game to collect human inputs on object
locations, followed by an accurate segmentation phase achieved by optimizing an
energy function encoding spatial and temporal constraints between object
regions as well as human-provided input. Performance analysis carried out on
challenging video datasets with some users playing the game demonstrated that
our method shows a better trade-off between annotation times and segmentation
accuracy than interactive video annotation and automated video object
segmentation approaches.
| [
{
"version": "v1",
"created": "Tue, 5 Jan 2016 13:48:05 GMT"
}
] | 2016-01-06T00:00:00 | [
[
"Palazzo",
"Simone",
""
],
[
"Spampinato",
"Concetto",
""
],
[
"Giordano",
"Daniela",
""
]
] | TITLE: Gamifying Video Object Segmentation
ABSTRACT: Video object segmentation can be considered as one of the most challenging
computer vision problems. Indeed, so far, no existing solution is able to
effectively deal with the peculiarities of real-world videos, especially in
cases of articulated motion and object occlusions; limitations that appear more
evident when we compare their performance with the human one. However, manually
segmenting objects in videos is largely impractical as it requires a lot of
human time and concentration. To address this problem, in this paper we propose
an interactive video object segmentation method, which exploits, on one hand,
the capability of humans to identify correctly objects in visual scenes, and on
the other hand, the collective human brainpower to solve challenging tasks. In
particular, our method relies on a web game to collect human inputs on object
locations, followed by an accurate segmentation phase achieved by optimizing an
energy function encoding spatial and temporal constraints between object
regions as well as human-provided input. Performance analysis carried out on
challenging video datasets with some users playing the game demonstrated that
our method shows a better trade-off between annotation times and segmentation
accuracy than interactive video annotation and automated video object
segmentation approaches.
| no_new_dataset | 0.947721 |
1502.04681 | Nitish Srivastava | Nitish Srivastava, Elman Mansimov and Ruslan Salakhutdinov | Unsupervised Learning of Video Representations using LSTMs | Added link to code on github | null | null | null | cs.LG cs.CV cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We use multilayer Long Short Term Memory (LSTM) networks to learn
representations of video sequences. Our model uses an encoder LSTM to map an
input sequence into a fixed length representation. This representation is
decoded using single or multiple decoder LSTMs to perform different tasks, such
as reconstructing the input sequence, or predicting the future sequence. We
experiment with two kinds of input sequences - patches of image pixels and
high-level representations ("percepts") of video frames extracted using a
pretrained convolutional net. We explore different design choices such as
whether the decoder LSTMs should condition on the generated output. We analyze
the outputs of the model qualitatively to see how well the model can
extrapolate the learned video representation into the future and into the past.
We try to visualize and interpret the learned features. We stress test the
model by running it on longer time scales and on out-of-domain data. We further
evaluate the representations by finetuning them for a supervised learning
problem - human action recognition on the UCF-101 and HMDB-51 datasets. We show
that the representations help improve classification accuracy, especially when
there are only a few training examples. Even models pretrained on unrelated
datasets (300 hours of YouTube videos) can help action recognition performance.
| [
{
"version": "v1",
"created": "Mon, 16 Feb 2015 20:00:07 GMT"
},
{
"version": "v2",
"created": "Tue, 31 Mar 2015 23:45:59 GMT"
},
{
"version": "v3",
"created": "Mon, 4 Jan 2016 00:42:07 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Srivastava",
"Nitish",
""
],
[
"Mansimov",
"Elman",
""
],
[
"Salakhutdinov",
"Ruslan",
""
]
] | TITLE: Unsupervised Learning of Video Representations using LSTMs
ABSTRACT: We use multilayer Long Short Term Memory (LSTM) networks to learn
representations of video sequences. Our model uses an encoder LSTM to map an
input sequence into a fixed length representation. This representation is
decoded using single or multiple decoder LSTMs to perform different tasks, such
as reconstructing the input sequence, or predicting the future sequence. We
experiment with two kinds of input sequences - patches of image pixels and
high-level representations ("percepts") of video frames extracted using a
pretrained convolutional net. We explore different design choices such as
whether the decoder LSTMs should condition on the generated output. We analyze
the outputs of the model qualitatively to see how well the model can
extrapolate the learned video representation into the future and into the past.
We try to visualize and interpret the learned features. We stress test the
model by running it on longer time scales and on out-of-domain data. We further
evaluate the representations by finetuning them for a supervised learning
problem - human action recognition on the UCF-101 and HMDB-51 datasets. We show
that the representations help improve classification accuracy, especially when
there are only a few training examples. Even models pretrained on unrelated
datasets (300 hours of YouTube videos) can help action recognition performance.
| no_new_dataset | 0.934873 |
1511.06406 | Daniel Jiwoong Im | Daniel Jiwoong Im, Sungjin Ahn, Roland Memisevic, Yoshua Bengio | Denoising Criterion for Variational Auto-Encoding Framework | ICLR conference submission | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Denoising autoencoders (DAE) are trained to reconstruct their clean inputs
with noise injected at the input level, while variational autoencoders (VAE)
are trained with noise injected in their stochastic hidden layer, with a
regularizer that encourages this noise injection. In this paper, we show that
injecting noise both in input and in the stochastic hidden layer can be
advantageous and we propose a modified variational lower bound as an improved
objective function in this setup. When input is corrupted, then the standard
VAE lower bound involves marginalizing the encoder conditional distribution
over the input noise, which makes the training criterion intractable. Instead,
we propose a modified training criterion which corresponds to a tractable bound
when input is corrupted. Experimentally, we find that the proposed denoising
variational autoencoder (DVAE) yields better average log-likelihood than the
VAE and the importance weighted autoencoder on the MNIST and Frey Face
datasets.
| [
{
"version": "v1",
"created": "Thu, 19 Nov 2015 21:56:21 GMT"
},
{
"version": "v2",
"created": "Mon, 4 Jan 2016 15:12:46 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Im",
"Daniel Jiwoong",
""
],
[
"Ahn",
"Sungjin",
""
],
[
"Memisevic",
"Roland",
""
],
[
"Bengio",
"Yoshua",
""
]
] | TITLE: Denoising Criterion for Variational Auto-Encoding Framework
ABSTRACT: Denoising autoencoders (DAE) are trained to reconstruct their clean inputs
with noise injected at the input level, while variational autoencoders (VAE)
are trained with noise injected in their stochastic hidden layer, with a
regularizer that encourages this noise injection. In this paper, we show that
injecting noise both in input and in the stochastic hidden layer can be
advantageous and we propose a modified variational lower bound as an improved
objective function in this setup. When input is corrupted, then the standard
VAE lower bound involves marginalizing the encoder conditional distribution
over the input noise, which makes the training criterion intractable. Instead,
we propose a modified training criterion which corresponds to a tractable bound
when input is corrupted. Experimentally, we find that the proposed denoising
variational autoencoder (DVAE) yields better average log-likelihood than the
VAE and the importance weighted autoencoder on the MNIST and Frey Face
datasets.
| no_new_dataset | 0.947672 |
1511.07425 | Mohammad Sabokrou | Mohammad Sabokrou, Mahmood Fathy, Mojtaba Hosseini | Real-Time Anomalous Behavior Detection and Localization in Crowded
Scenes | This paper has been withdrawn by the author due to some error in
experimental result. There are some mistakes | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | In this paper, we propose an accurate and real-time anomaly detection and
localization in crowded scenes, and two descriptors for representing anomalous
behavior in video are proposed. We consider a video as being a set of cubic
patches. Based on the low likelihood of an anomaly occurrence, and the
redundancy of structures in normal patches in videos, two (global and local)
views are considered for modeling the video. Our algorithm has two components,
for (1) representing the patches using local and global descriptors, and for
(2) modeling the training patches using a new representation. We have two
Gaussian models for all training patches respect to global and local
descriptors. The local and global features are based on structure similarity
between adjacent patches and the features that are learned in an unsupervised
way. We propose a fusion strategy to combine the two descriptors as the output
of our system. Experimental results show that our algorithm performs like a
state-of-the-art method on several standard datasets, but even is more
time-efficient.
| [
{
"version": "v1",
"created": "Sat, 21 Nov 2015 22:42:53 GMT"
},
{
"version": "v2",
"created": "Sat, 2 Jan 2016 06:10:47 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Sabokrou",
"Mohammad",
""
],
[
"Fathy",
"Mahmood",
""
],
[
"Hosseini",
"Mojtaba",
""
]
] | TITLE: Real-Time Anomalous Behavior Detection and Localization in Crowded
Scenes
ABSTRACT: In this paper, we propose an accurate and real-time anomaly detection and
localization in crowded scenes, and two descriptors for representing anomalous
behavior in video are proposed. We consider a video as being a set of cubic
patches. Based on the low likelihood of an anomaly occurrence, and the
redundancy of structures in normal patches in videos, two (global and local)
views are considered for modeling the video. Our algorithm has two components,
for (1) representing the patches using local and global descriptors, and for
(2) modeling the training patches using a new representation. We have two
Gaussian models for all training patches respect to global and local
descriptors. The local and global features are based on structure similarity
between adjacent patches and the features that are learned in an unsupervised
way. We propose a fusion strategy to combine the two descriptors as the output
of our system. Experimental results show that our algorithm performs like a
state-of-the-art method on several standard datasets, but even is more
time-efficient.
| no_new_dataset | 0.951908 |
1601.00024 | Ashish Sabharwal | Ashish Sabharwal, Horst Samulowitz, Gerald Tesauro | Selecting Near-Optimal Learners via Incremental Data Allocation | AAAI-2016: The Thirtieth AAAI Conference on Artificial Intelligence | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study a novel machine learning (ML) problem setting of sequentially
allocating small subsets of training data amongst a large set of classifiers.
The goal is to select a classifier that will give near-optimal accuracy when
trained on all data, while also minimizing the cost of misallocated samples.
This is motivated by large modern datasets and ML toolkits with many
combinations of learning algorithms and hyper-parameters. Inspired by the
principle of "optimism under uncertainty," we propose an innovative strategy,
Data Allocation using Upper Bounds (DAUB), which robustly achieves these
objectives across a variety of real-world datasets.
We further develop substantial theoretical support for DAUB in an idealized
setting where the expected accuracy of a classifier trained on $n$ samples can
be known exactly. Under these conditions we establish a rigorous sub-linear
bound on the regret of the approach (in terms of misallocated data), as well as
a rigorous bound on suboptimality of the selected classifier. Our accuracy
estimates using real-world datasets only entail mild violations of the
theoretical scenario, suggesting that the practical behavior of DAUB is likely
to approach the idealized behavior.
| [
{
"version": "v1",
"created": "Thu, 31 Dec 2015 22:19:09 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Sabharwal",
"Ashish",
""
],
[
"Samulowitz",
"Horst",
""
],
[
"Tesauro",
"Gerald",
""
]
] | TITLE: Selecting Near-Optimal Learners via Incremental Data Allocation
ABSTRACT: We study a novel machine learning (ML) problem setting of sequentially
allocating small subsets of training data amongst a large set of classifiers.
The goal is to select a classifier that will give near-optimal accuracy when
trained on all data, while also minimizing the cost of misallocated samples.
This is motivated by large modern datasets and ML toolkits with many
combinations of learning algorithms and hyper-parameters. Inspired by the
principle of "optimism under uncertainty," we propose an innovative strategy,
Data Allocation using Upper Bounds (DAUB), which robustly achieves these
objectives across a variety of real-world datasets.
We further develop substantial theoretical support for DAUB in an idealized
setting where the expected accuracy of a classifier trained on $n$ samples can
be known exactly. Under these conditions we establish a rigorous sub-linear
bound on the regret of the approach (in terms of misallocated data), as well as
a rigorous bound on suboptimality of the selected classifier. Our accuracy
estimates using real-world datasets only entail mild violations of the
theoretical scenario, suggesting that the practical behavior of DAUB is likely
to approach the idealized behavior.
| no_new_dataset | 0.94256 |
1601.00073 | Oliver Kennedy | Arindam Nandi, Ying Yang, Oliver Kennedy, Boris Glavic, Ronny Fehling,
Zhen Hua Liu, Dieter Gawlick | Mimir: Bringing CTables into Practice | Under submission; The first two authors should be considered a joint
first-author | null | null | null | cs.DB cs.PL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The present state of the art in analytics requires high upfront investment of
human effort and computational resources to curate datasets, even before the
first query is posed. So-called pay-as-you-go data curation techniques allow
these high costs to be spread out, first by enabling queries over uncertain and
incomplete data, and then by assessing the quality of the query results. We
describe the design of a system, called Mimir, around a recently introduced
class of probabilistic pay-as-you-go data cleaning operators called Lenses.
Mimir wraps around any deterministic database engine using JDBC, extending it
with support for probabilistic query processing. Queries processed through
Mimir produce uncertainty-annotated result cursors that allow client
applications to quickly assess result quality and provenance. We also present a
GUI that provides analysts with an interactive tool for exploring the
uncertainty exposed by the system. Finally, we present optimizations that make
Lenses scalable, and validate this claim through experimental evidence.
| [
{
"version": "v1",
"created": "Fri, 1 Jan 2016 11:21:33 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Nandi",
"Arindam",
""
],
[
"Yang",
"Ying",
""
],
[
"Kennedy",
"Oliver",
""
],
[
"Glavic",
"Boris",
""
],
[
"Fehling",
"Ronny",
""
],
[
"Liu",
"Zhen Hua",
""
],
[
"Gawlick",
"Dieter",
""
]
] | TITLE: Mimir: Bringing CTables into Practice
ABSTRACT: The present state of the art in analytics requires high upfront investment of
human effort and computational resources to curate datasets, even before the
first query is posed. So-called pay-as-you-go data curation techniques allow
these high costs to be spread out, first by enabling queries over uncertain and
incomplete data, and then by assessing the quality of the query results. We
describe the design of a system, called Mimir, around a recently introduced
class of probabilistic pay-as-you-go data cleaning operators called Lenses.
Mimir wraps around any deterministic database engine using JDBC, extending it
with support for probabilistic query processing. Queries processed through
Mimir produce uncertainty-annotated result cursors that allow client
applications to quickly assess result quality and provenance. We also present a
GUI that provides analysts with an interactive tool for exploring the
uncertainty exposed by the system. Finally, we present optimizations that make
Lenses scalable, and validate this claim through experimental evidence.
| no_new_dataset | 0.94428 |
1601.00236 | Chetan Tonde | Praneeth Vepakomma and Chetan Tonde and Ahmed Elgammal | Supervised Dimensionality Reduction via Distance Correlation
Maximization | 23 pages, 6 figures | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In our work, we propose a novel formulation for supervised dimensionality
reduction based on a nonlinear dependency criterion called Statistical Distance
Correlation, Szekely et. al. (2007). We propose an objective which is free of
distributional assumptions on regression variables and regression model
assumptions. Our proposed formulation is based on learning a low-dimensional
feature representation $\mathbf{z}$, which maximizes the squared sum of
Distance Correlations between low dimensional features $\mathbf{z}$ and
response $y$, and also between features $\mathbf{z}$ and covariates
$\mathbf{x}$. We propose a novel algorithm to optimize our proposed objective
using the Generalized Minimization Maximizaiton method of \Parizi et. al.
(2015). We show superior empirical results on multiple datasets proving the
effectiveness of our proposed approach over several relevant state-of-the-art
supervised dimensionality reduction methods.
| [
{
"version": "v1",
"created": "Sun, 3 Jan 2016 00:14:23 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Vepakomma",
"Praneeth",
""
],
[
"Tonde",
"Chetan",
""
],
[
"Elgammal",
"Ahmed",
""
]
] | TITLE: Supervised Dimensionality Reduction via Distance Correlation
Maximization
ABSTRACT: In our work, we propose a novel formulation for supervised dimensionality
reduction based on a nonlinear dependency criterion called Statistical Distance
Correlation, Szekely et. al. (2007). We propose an objective which is free of
distributional assumptions on regression variables and regression model
assumptions. Our proposed formulation is based on learning a low-dimensional
feature representation $\mathbf{z}$, which maximizes the squared sum of
Distance Correlations between low dimensional features $\mathbf{z}$ and
response $y$, and also between features $\mathbf{z}$ and covariates
$\mathbf{x}$. We propose a novel algorithm to optimize our proposed objective
using the Generalized Minimization Maximizaiton method of \Parizi et. al.
(2015). We show superior empirical results on multiple datasets proving the
effectiveness of our proposed approach over several relevant state-of-the-art
supervised dimensionality reduction methods.
| no_new_dataset | 0.946695 |
1601.00400 | Abrar Abdulnabi | Abrar H. Abdulnabi, Gang Wang, Jiwen Lu, Kui Jia | Multi-task CNN Model for Attribute Prediction | 11 pages, 3 figures, ieee transaction paper | IEEE Transactions on Multimedia, Nov 2015, pp. 1949-1959 | 10.1109/TMM.2015.2477680 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes a joint multi-task learning algorithm to better predict
attributes in images using deep convolutional neural networks (CNN). We
consider learning binary semantic attributes through a multi-task CNN model,
where each CNN will predict one binary attribute. The multi-task learning
allows CNN models to simultaneously share visual knowledge among different
attribute categories. Each CNN will generate attribute-specific feature
representations, and then we apply multi-task learning on the features to
predict their attributes. In our multi-task framework, we propose a method to
decompose the overall model's parameters into a latent task matrix and
combination matrix. Furthermore, under-sampled classifiers can leverage shared
statistics from other classifiers to improve their performance. Natural
grouping of attributes is applied such that attributes in the same group are
encouraged to share more knowledge. Meanwhile, attributes in different groups
will generally compete with each other, and consequently share less knowledge.
We show the effectiveness of our method on two popular attribute datasets.
| [
{
"version": "v1",
"created": "Mon, 4 Jan 2016 07:42:56 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Abdulnabi",
"Abrar H.",
""
],
[
"Wang",
"Gang",
""
],
[
"Lu",
"Jiwen",
""
],
[
"Jia",
"Kui",
""
]
] | TITLE: Multi-task CNN Model for Attribute Prediction
ABSTRACT: This paper proposes a joint multi-task learning algorithm to better predict
attributes in images using deep convolutional neural networks (CNN). We
consider learning binary semantic attributes through a multi-task CNN model,
where each CNN will predict one binary attribute. The multi-task learning
allows CNN models to simultaneously share visual knowledge among different
attribute categories. Each CNN will generate attribute-specific feature
representations, and then we apply multi-task learning on the features to
predict their attributes. In our multi-task framework, we propose a method to
decompose the overall model's parameters into a latent task matrix and
combination matrix. Furthermore, under-sampled classifiers can leverage shared
statistics from other classifiers to improve their performance. Natural
grouping of attributes is applied such that attributes in the same group are
encouraged to share more knowledge. Meanwhile, attributes in different groups
will generally compete with each other, and consequently share less knowledge.
We show the effectiveness of our method on two popular attribute datasets.
| no_new_dataset | 0.944791 |
1601.00626 | Tim Weninger PhD | Baoxu Shi and Tim Weninger | Scalable Models for Computing Hierarchies in Information Networks | Preprint for "Knowledge and Information Systems" paper, in press | null | null | null | cs.AI cs.DL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Information hierarchies are organizational structures that often used to
organize and present large and complex information as well as provide a
mechanism for effective human navigation. Fortunately, many statistical and
computational models exist that automatically generate hierarchies; however,
the existing approaches do not consider linkages in information {\em networks}
that are increasingly common in real-world scenarios. Current approaches also
tend to present topics as an abstract probably distribution over words, etc
rather than as tangible nodes from the original network. Furthermore, the
statistical techniques present in many previous works are not yet capable of
processing data at Web-scale. In this paper we present the Hierarchical
Document Topic Model (HDTM), which uses a distributed vertex-programming
process to calculate a nonparametric Bayesian generative model. Experiments on
three medium size data sets and the entire Wikipedia dataset show that HDTM can
infer accurate hierarchies even over large information networks.
| [
{
"version": "v1",
"created": "Mon, 4 Jan 2016 20:05:19 GMT"
}
] | 2016-01-05T00:00:00 | [
[
"Shi",
"Baoxu",
""
],
[
"Weninger",
"Tim",
""
]
] | TITLE: Scalable Models for Computing Hierarchies in Information Networks
ABSTRACT: Information hierarchies are organizational structures that often used to
organize and present large and complex information as well as provide a
mechanism for effective human navigation. Fortunately, many statistical and
computational models exist that automatically generate hierarchies; however,
the existing approaches do not consider linkages in information {\em networks}
that are increasingly common in real-world scenarios. Current approaches also
tend to present topics as an abstract probably distribution over words, etc
rather than as tangible nodes from the original network. Furthermore, the
statistical techniques present in many previous works are not yet capable of
processing data at Web-scale. In this paper we present the Hierarchical
Document Topic Model (HDTM), which uses a distributed vertex-programming
process to calculate a nonparametric Bayesian generative model. Experiments on
three medium size data sets and the entire Wikipedia dataset show that HDTM can
infer accurate hierarchies even over large information networks.
| no_new_dataset | 0.947769 |
1403.5864 | Ying Long | Ying Long and Yao Shen | Mapping parcel-level urban areas for a large geographical area | 21 pages, 9 figures, 3 tables | null | 10.1080/00045608.2015.1095062 | null | cs.OH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As a vital indicator for measuring urban development, urban areas are
expected to be identified explicitly and conveniently with widely available
dataset thereby benefiting the planning decisions and relevant urban studies.
Existing approaches to identify urban areas normally based on mid-resolution
sensing dataset, socioeconomic information (e.g. population density) generally
associate with low-resolution in space, e.g. cells with several square
kilometers or even larger towns/wards. Yet, few of them pay attention to
defining urban areas with micro data in a fine-scaled manner with large extend
scale by incorporating the morphological and functional characteristics. This
paper investigates an automated framework to delineate urban areas in the
parcel level, using increasingly available ordnance surveys for generating all
parcels (or geo-units) and ubiquitous points of interest (POIs) for inferring
density of each parcel. A vector cellular automata model was adopted for
identifying urban parcels from all generated parcels, taking into account
density, neighborhood condition, and other spatial variables of each parcel. We
applied this approach for mapping urban areas of all 654 Chinese cities and
compared them with those interpreted from mid-resolution remote sensing images
and inferred by population density and road intersections. Our proposed
framework is proved to be more straight-forward, time-saving and fine-scaled,
compared with other existing ones, and reclaim the need for consistency,
efficiency and availability in defining urban areas with well-consideration of
omnipresent spatial and functional factors across cities.
| [
{
"version": "v1",
"created": "Mon, 24 Mar 2014 06:39:17 GMT"
}
] | 2016-01-01T00:00:00 | [
[
"Long",
"Ying",
""
],
[
"Shen",
"Yao",
""
]
] | TITLE: Mapping parcel-level urban areas for a large geographical area
ABSTRACT: As a vital indicator for measuring urban development, urban areas are
expected to be identified explicitly and conveniently with widely available
dataset thereby benefiting the planning decisions and relevant urban studies.
Existing approaches to identify urban areas normally based on mid-resolution
sensing dataset, socioeconomic information (e.g. population density) generally
associate with low-resolution in space, e.g. cells with several square
kilometers or even larger towns/wards. Yet, few of them pay attention to
defining urban areas with micro data in a fine-scaled manner with large extend
scale by incorporating the morphological and functional characteristics. This
paper investigates an automated framework to delineate urban areas in the
parcel level, using increasingly available ordnance surveys for generating all
parcels (or geo-units) and ubiquitous points of interest (POIs) for inferring
density of each parcel. A vector cellular automata model was adopted for
identifying urban parcels from all generated parcels, taking into account
density, neighborhood condition, and other spatial variables of each parcel. We
applied this approach for mapping urban areas of all 654 Chinese cities and
compared them with those interpreted from mid-resolution remote sensing images
and inferred by population density and road intersections. Our proposed
framework is proved to be more straight-forward, time-saving and fine-scaled,
compared with other existing ones, and reclaim the need for consistency,
efficiency and availability in defining urban areas with well-consideration of
omnipresent spatial and functional factors across cities.
| no_new_dataset | 0.953751 |
1512.09295 | Qirong Ho | Eric P. Xing, Qirong Ho, Pengtao Xie, Wei Dai | Strategies and Principles of Distributed Machine Learning on Big Data | null | null | null | null | stat.ML cs.DC cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rise of Big Data has led to new demands for Machine Learning (ML) systems
to learn complex models with millions to billions of parameters, that promise
adequate capacity to digest massive datasets and offer powerful predictive
analytics thereupon. In order to run ML algorithms at such scales, on a
distributed cluster with 10s to 1000s of machines, it is often the case that
significant engineering efforts are required --- and one might fairly ask if
such engineering truly falls within the domain of ML research or not. Taking
the view that Big ML systems can benefit greatly from ML-rooted statistical and
algorithmic insights --- and that ML researchers should therefore not shy away
from such systems design --- we discuss a series of principles and strategies
distilled from our recent efforts on industrial-scale ML solutions. These
principles and strategies span a continuum from application, to engineering,
and to theoretical research and development of Big ML systems and
architectures, with the goal of understanding how to make them efficient,
generally-applicable, and supported with convergence and scaling guarantees.
They concern four key questions which traditionally receive little attention in
ML research: How to distribute an ML program over a cluster? How to bridge ML
computation with inter-machine communication? How to perform such
communication? What should be communicated between machines? By exposing
underlying statistical and algorithmic characteristics unique to ML programs
but not typically seen in traditional computer programs, and by dissecting
successful cases to reveal how we have harnessed these principles to design and
develop both high-performance distributed ML software as well as
general-purpose ML frameworks, we present opportunities for ML researchers and
practitioners to further shape and grow the area that lies between ML and
systems.
| [
{
"version": "v1",
"created": "Thu, 31 Dec 2015 14:33:53 GMT"
}
] | 2016-01-01T00:00:00 | [
[
"Xing",
"Eric P.",
""
],
[
"Ho",
"Qirong",
""
],
[
"Xie",
"Pengtao",
""
],
[
"Dai",
"Wei",
""
]
] | TITLE: Strategies and Principles of Distributed Machine Learning on Big Data
ABSTRACT: The rise of Big Data has led to new demands for Machine Learning (ML) systems
to learn complex models with millions to billions of parameters, that promise
adequate capacity to digest massive datasets and offer powerful predictive
analytics thereupon. In order to run ML algorithms at such scales, on a
distributed cluster with 10s to 1000s of machines, it is often the case that
significant engineering efforts are required --- and one might fairly ask if
such engineering truly falls within the domain of ML research or not. Taking
the view that Big ML systems can benefit greatly from ML-rooted statistical and
algorithmic insights --- and that ML researchers should therefore not shy away
from such systems design --- we discuss a series of principles and strategies
distilled from our recent efforts on industrial-scale ML solutions. These
principles and strategies span a continuum from application, to engineering,
and to theoretical research and development of Big ML systems and
architectures, with the goal of understanding how to make them efficient,
generally-applicable, and supported with convergence and scaling guarantees.
They concern four key questions which traditionally receive little attention in
ML research: How to distribute an ML program over a cluster? How to bridge ML
computation with inter-machine communication? How to perform such
communication? What should be communicated between machines? By exposing
underlying statistical and algorithmic characteristics unique to ML programs
but not typically seen in traditional computer programs, and by dissecting
successful cases to reveal how we have harnessed these principles to design and
develop both high-performance distributed ML software as well as
general-purpose ML frameworks, we present opportunities for ML researchers and
practitioners to further shape and grow the area that lies between ML and
systems.
| no_new_dataset | 0.937383 |
1506.05439 | Charlie Frogner | Charlie Frogner, Chiyuan Zhang, Hossein Mobahi, Mauricio Araya-Polo,
Tomaso Poggio | Learning with a Wasserstein Loss | NIPS 2015; v3 updates Algorithm 1 and Equations 6, 8 | null | null | null | cs.LG cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning to predict multi-label outputs is challenging, but in many problems
there is a natural metric on the outputs that can be used to improve
predictions. In this paper we develop a loss function for multi-label learning,
based on the Wasserstein distance. The Wasserstein distance provides a natural
notion of dissimilarity for probability measures. Although optimizing with
respect to the exact Wasserstein distance is costly, recent work has described
a regularized approximation that is efficiently computed. We describe an
efficient learning algorithm based on this regularization, as well as a novel
extension of the Wasserstein distance from probability measures to unnormalized
measures. We also describe a statistical learning bound for the loss. The
Wasserstein loss can encourage smoothness of the predictions with respect to a
chosen metric on the output space. We demonstrate this property on a real-data
tag prediction problem, using the Yahoo Flickr Creative Commons dataset,
outperforming a baseline that doesn't use the metric.
| [
{
"version": "v1",
"created": "Wed, 17 Jun 2015 19:36:41 GMT"
},
{
"version": "v2",
"created": "Fri, 6 Nov 2015 03:46:05 GMT"
},
{
"version": "v3",
"created": "Wed, 30 Dec 2015 01:08:11 GMT"
}
] | 2015-12-31T00:00:00 | [
[
"Frogner",
"Charlie",
""
],
[
"Zhang",
"Chiyuan",
""
],
[
"Mobahi",
"Hossein",
""
],
[
"Araya-Polo",
"Mauricio",
""
],
[
"Poggio",
"Tomaso",
""
]
] | TITLE: Learning with a Wasserstein Loss
ABSTRACT: Learning to predict multi-label outputs is challenging, but in many problems
there is a natural metric on the outputs that can be used to improve
predictions. In this paper we develop a loss function for multi-label learning,
based on the Wasserstein distance. The Wasserstein distance provides a natural
notion of dissimilarity for probability measures. Although optimizing with
respect to the exact Wasserstein distance is costly, recent work has described
a regularized approximation that is efficiently computed. We describe an
efficient learning algorithm based on this regularization, as well as a novel
extension of the Wasserstein distance from probability measures to unnormalized
measures. We also describe a statistical learning bound for the loss. The
Wasserstein loss can encourage smoothness of the predictions with respect to a
chosen metric on the output space. We demonstrate this property on a real-data
tag prediction problem, using the Yahoo Flickr Creative Commons dataset,
outperforming a baseline that doesn't use the metric.
| no_new_dataset | 0.947721 |
1512.08669 | Da-Han Wang | Da-Han Wang, Hanzi Wang, Dong Zhang, Jonathan Li, David Zhang | Robust Scene Text Recognition Using Sparse Coding based Features | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose an effective scene text recognition method using
sparse coding based features, called Histograms of Sparse Codes (HSC) features.
For character detection, we use the HSC features instead of using the
Histograms of Oriented Gradients (HOG) features. The HSC features are extracted
by computing sparse codes with dictionaries that are learned from data using
K-SVD, and aggregating per-pixel sparse codes to form local histograms. For
word recognition, we integrate multiple cues including character detection
scores and geometric contexts in an objective function. The final recognition
results are obtained by searching for the words which correspond to the maximum
value of the objective function. The parameters in the objective function are
learned using the Minimum Classification Error (MCE) training method.
Experiments on several challenging datasets demonstrate that the proposed
HSC-based scene text recognition method outperforms HOG-based methods
significantly and outperforms most state-of-the-art methods.
| [
{
"version": "v1",
"created": "Tue, 29 Dec 2015 12:50:40 GMT"
}
] | 2015-12-31T00:00:00 | [
[
"Wang",
"Da-Han",
""
],
[
"Wang",
"Hanzi",
""
],
[
"Zhang",
"Dong",
""
],
[
"Li",
"Jonathan",
""
],
[
"Zhang",
"David",
""
]
] | TITLE: Robust Scene Text Recognition Using Sparse Coding based Features
ABSTRACT: In this paper, we propose an effective scene text recognition method using
sparse coding based features, called Histograms of Sparse Codes (HSC) features.
For character detection, we use the HSC features instead of using the
Histograms of Oriented Gradients (HOG) features. The HSC features are extracted
by computing sparse codes with dictionaries that are learned from data using
K-SVD, and aggregating per-pixel sparse codes to form local histograms. For
word recognition, we integrate multiple cues including character detection
scores and geometric contexts in an objective function. The final recognition
results are obtained by searching for the words which correspond to the maximum
value of the objective function. The parameters in the objective function are
learned using the Minimum Classification Error (MCE) training method.
Experiments on several challenging datasets demonstrate that the proposed
HSC-based scene text recognition method outperforms HOG-based methods
significantly and outperforms most state-of-the-art methods.
| no_new_dataset | 0.947137 |
1512.08787 | Ravi Ganti | Ravi Ganti, Laura Balzano, Rebecca Willett | Matrix Completion Under Monotonic Single Index Models | 21 pages, 5 figures, 1 table. Accepted for publication at NIPS 2015 | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most recent results in matrix completion assume that the matrix under
consideration is low-rank or that the columns are in a union of low-rank
subspaces. In real-world settings, however, the linear structure underlying
these models is distorted by a (typically unknown) nonlinear transformation.
This paper addresses the challenge of matrix completion in the face of such
nonlinearities. Given a few observations of a matrix that are obtained by
applying a Lipschitz, monotonic function to a low rank matrix, our task is to
estimate the remaining unobserved entries. We propose a novel matrix completion
method that alternates between low-rank matrix estimation and monotonic
function estimation to estimate the missing matrix elements. Mean squared error
bounds provide insight into how well the matrix can be estimated based on the
size, rank of the matrix and properties of the nonlinear transformation.
Empirical results on synthetic and real-world datasets demonstrate the
competitiveness of the proposed approach.
| [
{
"version": "v1",
"created": "Tue, 29 Dec 2015 20:52:41 GMT"
}
] | 2015-12-31T00:00:00 | [
[
"Ganti",
"Ravi",
""
],
[
"Balzano",
"Laura",
""
],
[
"Willett",
"Rebecca",
""
]
] | TITLE: Matrix Completion Under Monotonic Single Index Models
ABSTRACT: Most recent results in matrix completion assume that the matrix under
consideration is low-rank or that the columns are in a union of low-rank
subspaces. In real-world settings, however, the linear structure underlying
these models is distorted by a (typically unknown) nonlinear transformation.
This paper addresses the challenge of matrix completion in the face of such
nonlinearities. Given a few observations of a matrix that are obtained by
applying a Lipschitz, monotonic function to a low rank matrix, our task is to
estimate the remaining unobserved entries. We propose a novel matrix completion
method that alternates between low-rank matrix estimation and monotonic
function estimation to estimate the missing matrix elements. Mean squared error
bounds provide insight into how well the matrix can be estimated based on the
size, rank of the matrix and properties of the nonlinear transformation.
Empirical results on synthetic and real-world datasets demonstrate the
competitiveness of the proposed approach.
| no_new_dataset | 0.94474 |
1512.08799 | Hao Wu | Hao Wu, Maoyuan Sun, Peng Mi, Nikolaj Tatti, Chris North, Naren
Ramakrishnan | Interactive Discovery of Coordinated Relationship Chains with Maximum
Entropy Models | The journal version of paper is submitted for publication | null | null | null | cs.DB cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Modern visual analytic tools promote human-in-the-loop analysis but are
limited in their ability to direct the user toward interesting and promising
directions of study. This problem is especially acute when the analysis task is
exploratory in nature, e.g., the discovery of potentially coordinated
relationships in massive text datasets. Such tasks are very common in domains
like intelligence analysis and security forensics where the goal is to uncover
surprising coalitions bridging multiple types of relations. We introduce new
maximum entropy models to discover surprising chains of relationships
leveraging count data about entity occurrences in documents. These models are
embedded in a visual analytic system called MERCER that treats relationship
bundles as first class objects and directs the user toward promising lines of
inquiry. We demonstrate how user input can judiciously direct analysis toward
valid conclusions whereas a purely algorithmic approach could be led astray.
Experimental results on both synthetic and real datasets from the intelligence
community are presented.
| [
{
"version": "v1",
"created": "Tue, 29 Dec 2015 21:27:05 GMT"
}
] | 2015-12-31T00:00:00 | [
[
"Wu",
"Hao",
""
],
[
"Sun",
"Maoyuan",
""
],
[
"Mi",
"Peng",
""
],
[
"Tatti",
"Nikolaj",
""
],
[
"North",
"Chris",
""
],
[
"Ramakrishnan",
"Naren",
""
]
] | TITLE: Interactive Discovery of Coordinated Relationship Chains with Maximum
Entropy Models
ABSTRACT: Modern visual analytic tools promote human-in-the-loop analysis but are
limited in their ability to direct the user toward interesting and promising
directions of study. This problem is especially acute when the analysis task is
exploratory in nature, e.g., the discovery of potentially coordinated
relationships in massive text datasets. Such tasks are very common in domains
like intelligence analysis and security forensics where the goal is to uncover
surprising coalitions bridging multiple types of relations. We introduce new
maximum entropy models to discover surprising chains of relationships
leveraging count data about entity occurrences in documents. These models are
embedded in a visual analytic system called MERCER that treats relationship
bundles as first class objects and directs the user toward promising lines of
inquiry. We demonstrate how user input can judiciously direct analysis toward
valid conclusions whereas a purely algorithmic approach could be led astray.
Experimental results on both synthetic and real datasets from the intelligence
community are presented.
| no_new_dataset | 0.951006 |
1512.08826 | Manfred Lau | Kapil Dev, Manfred Lau | Improving Style Similarity Metrics of 3D Shapes | null | null | null | null | cs.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The idea of style similarity metrics has been recently developed for various
media types such as 2D clip art and 3D shapes. We explore this style metric
problem and improve existing style similarity metrics of 3D shapes in four
novel ways. First, we consider the color and texture of 3D shapes which are
important properties that have not been previously considered. Second, we
explore the effect of clustering a dataset of 3D models by comparing between
style metrics for a single object type and style metrics that combine clusters
of object types. Third, we explore the idea of user-guided learning for this
problem. Fourth, we introduce an iterative approach that can learn a metric
from a general set of 3D models. We demonstrate these contributions with
various classes of 3D shapes and with applications such as style-based
similarity search and scene composition.
| [
{
"version": "v1",
"created": "Wed, 30 Dec 2015 02:26:46 GMT"
}
] | 2015-12-31T00:00:00 | [
[
"Dev",
"Kapil",
""
],
[
"Lau",
"Manfred",
""
]
] | TITLE: Improving Style Similarity Metrics of 3D Shapes
ABSTRACT: The idea of style similarity metrics has been recently developed for various
media types such as 2D clip art and 3D shapes. We explore this style metric
problem and improve existing style similarity metrics of 3D shapes in four
novel ways. First, we consider the color and texture of 3D shapes which are
important properties that have not been previously considered. Second, we
explore the effect of clustering a dataset of 3D models by comparing between
style metrics for a single object type and style metrics that combine clusters
of object types. Third, we explore the idea of user-guided learning for this
problem. Fourth, we introduce an iterative approach that can learn a metric
from a general set of 3D models. We demonstrate these contributions with
various classes of 3D shapes and with applications such as style-based
similarity search and scene composition.
| no_new_dataset | 0.952442 |
1512.09041 | Chenliang Xu | Chenliang Xu and Jason J. Corso | Actor-Action Semantic Segmentation with Grouping Process Models | Technical report | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Actor-action semantic segmentation made an important step toward advanced
video understanding problems: what action is happening; who is performing the
action; and where is the action in space-time. Current models for this problem
are local, based on layered CRFs, and are unable to capture long-ranging
interaction of video parts. We propose a new model that combines these local
labeling CRFs with a hierarchical supervoxel decomposition. The supervoxels
provide cues for possible groupings of nodes, at various scales, in the CRFs to
encourage adaptive, high-order groups for more effective labeling. Our model is
dynamic and continuously exchanges information during inference: the local CRFs
influence what supervoxels in the hierarchy are active, and these active nodes
influence the connectivity in the CRF; we hence call it a grouping process
model. The experimental results on a recent large-scale video dataset show a
large margin of 60% relative improvement over the state of the art, which
demonstrates the effectiveness of the dynamic, bidirectional flow between
labeling and grouping.
| [
{
"version": "v1",
"created": "Wed, 30 Dec 2015 18:07:45 GMT"
}
] | 2015-12-31T00:00:00 | [
[
"Xu",
"Chenliang",
""
],
[
"Corso",
"Jason J.",
""
]
] | TITLE: Actor-Action Semantic Segmentation with Grouping Process Models
ABSTRACT: Actor-action semantic segmentation made an important step toward advanced
video understanding problems: what action is happening; who is performing the
action; and where is the action in space-time. Current models for this problem
are local, based on layered CRFs, and are unable to capture long-ranging
interaction of video parts. We propose a new model that combines these local
labeling CRFs with a hierarchical supervoxel decomposition. The supervoxels
provide cues for possible groupings of nodes, at various scales, in the CRFs to
encourage adaptive, high-order groups for more effective labeling. Our model is
dynamic and continuously exchanges information during inference: the local CRFs
influence what supervoxels in the hierarchy are active, and these active nodes
influence the connectivity in the CRF; we hence call it a grouping process
model. The experimental results on a recent large-scale video dataset show a
large margin of 60% relative improvement over the state of the art, which
demonstrates the effectiveness of the dynamic, bidirectional flow between
labeling and grouping.
| no_new_dataset | 0.953232 |
1508.01006 | Dongxu Zhang | Dongxu Zhang and Dong Wang | Relation Classification via Recurrent Neural Network | null | null | null | null | cs.CL cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep learning has gained much success in sentence-level relation
classification. For example, convolutional neural networks (CNN) have delivered
competitive performance without much effort on feature engineering as the
conventional pattern-based methods. Thus a lot of works have been produced
based on CNN structures. However, a key issue that has not been well addressed
by the CNN-based method is the lack of capability to learn temporal features,
especially long-distance dependency between nominal pairs. In this paper, we
propose a simple framework based on recurrent neural networks (RNN) and compare
it with CNN-based model. To show the limitation of popular used SemEval-2010
Task 8 dataset, we introduce another dataset refined from MIMLRE(Angeli et al.,
2014). Experiments on two different datasets strongly indicates that the
RNN-based model can deliver better performance on relation classification, and
it is particularly capable of learning long-distance relation patterns. This
makes it suitable for real-world applications where complicated expressions are
often involved.
| [
{
"version": "v1",
"created": "Wed, 5 Aug 2015 09:03:46 GMT"
},
{
"version": "v2",
"created": "Fri, 25 Dec 2015 03:51:00 GMT"
}
] | 2015-12-29T00:00:00 | [
[
"Zhang",
"Dongxu",
""
],
[
"Wang",
"Dong",
""
]
] | TITLE: Relation Classification via Recurrent Neural Network
ABSTRACT: Deep learning has gained much success in sentence-level relation
classification. For example, convolutional neural networks (CNN) have delivered
competitive performance without much effort on feature engineering as the
conventional pattern-based methods. Thus a lot of works have been produced
based on CNN structures. However, a key issue that has not been well addressed
by the CNN-based method is the lack of capability to learn temporal features,
especially long-distance dependency between nominal pairs. In this paper, we
propose a simple framework based on recurrent neural networks (RNN) and compare
it with CNN-based model. To show the limitation of popular used SemEval-2010
Task 8 dataset, we introduce another dataset refined from MIMLRE(Angeli et al.,
2014). Experiments on two different datasets strongly indicates that the
RNN-based model can deliver better performance on relation classification, and
it is particularly capable of learning long-distance relation patterns. This
makes it suitable for real-world applications where complicated expressions are
often involved.
| new_dataset | 0.970799 |
1512.06915 | Yan Cui | Yan Cui | An Evaluation of Yelp Dataset | null | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Yelp is one of the largest online searching and reviewing systems for kinds
of businesses, including restaurants, shopping, home services et al. Analyzing
the real world data from Yelp is valuable in acquiring the interests of users,
which helps to improve the design of the next generation system. This paper
targets the evaluation of Yelp dataset, which is provided in the Yelp data
challenge. A bunch of interesting results are found. For instance, to reach any
one in the Yelp social network, one only needs 4.5 hops on average, which
verifies the classical six degree separation theory; Elite user mechanism is
especially effective in maintaining the healthy of the whole network; Users who
write less than 100 business reviews dominate. Those insights are expected to
be considered by Yelp to make intelligent business decisions in the future.
| [
{
"version": "v1",
"created": "Mon, 21 Dec 2015 23:54:08 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Dec 2015 23:24:39 GMT"
}
] | 2015-12-29T00:00:00 | [
[
"Cui",
"Yan",
""
]
] | TITLE: An Evaluation of Yelp Dataset
ABSTRACT: Yelp is one of the largest online searching and reviewing systems for kinds
of businesses, including restaurants, shopping, home services et al. Analyzing
the real world data from Yelp is valuable in acquiring the interests of users,
which helps to improve the design of the next generation system. This paper
targets the evaluation of Yelp dataset, which is provided in the Yelp data
challenge. A bunch of interesting results are found. For instance, to reach any
one in the Yelp social network, one only needs 4.5 hops on average, which
verifies the classical six degree separation theory; Elite user mechanism is
especially effective in maintaining the healthy of the whole network; Users who
write less than 100 business reviews dominate. Those insights are expected to
be considered by Yelp to make intelligent business decisions in the future.
| no_new_dataset | 0.94868 |
1512.07928 | Seunghoon Hong | Seunghoon Hong, Junhyuk Oh, Bohyung Han and Honglak Lee | Learning Transferrable Knowledge for Semantic Segmentation with Deep
Convolutional Neural Network | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a novel weakly-supervised semantic segmentation algorithm based on
Deep Convolutional Neural Network (DCNN). Contrary to existing
weakly-supervised approaches, our algorithm exploits auxiliary segmentation
annotations available for different categories to guide segmentations on images
with only image-level class labels. To make the segmentation knowledge
transferrable across categories, we design a decoupled encoder-decoder
architecture with attention model. In this architecture, the model generates
spatial highlights of each category presented in an image using an attention
model, and subsequently generates foreground segmentation for each highlighted
region using decoder. Combining attention model, we show that the decoder
trained with segmentation annotations in different categories can boost the
performance of weakly-supervised semantic segmentation. The proposed algorithm
demonstrates substantially improved performance compared to the
state-of-the-art weakly-supervised techniques in challenging PASCAL VOC 2012
dataset when our model is trained with the annotations in 60 exclusive
categories in Microsoft COCO dataset.
| [
{
"version": "v1",
"created": "Thu, 24 Dec 2015 22:33:27 GMT"
}
] | 2015-12-29T00:00:00 | [
[
"Hong",
"Seunghoon",
""
],
[
"Oh",
"Junhyuk",
""
],
[
"Han",
"Bohyung",
""
],
[
"Lee",
"Honglak",
""
]
] | TITLE: Learning Transferrable Knowledge for Semantic Segmentation with Deep
Convolutional Neural Network
ABSTRACT: We propose a novel weakly-supervised semantic segmentation algorithm based on
Deep Convolutional Neural Network (DCNN). Contrary to existing
weakly-supervised approaches, our algorithm exploits auxiliary segmentation
annotations available for different categories to guide segmentations on images
with only image-level class labels. To make the segmentation knowledge
transferrable across categories, we design a decoupled encoder-decoder
architecture with attention model. In this architecture, the model generates
spatial highlights of each category presented in an image using an attention
model, and subsequently generates foreground segmentation for each highlighted
region using decoder. Combining attention model, we show that the decoder
trained with segmentation annotations in different categories can boost the
performance of weakly-supervised semantic segmentation. The proposed algorithm
demonstrates substantially improved performance compared to the
state-of-the-art weakly-supervised techniques in challenging PASCAL VOC 2012
dataset when our model is trained with the annotations in 60 exclusive
categories in Microsoft COCO dataset.
| no_new_dataset | 0.952309 |
1512.07951 | M. Avendi | M. R. Avendi, A. Kheradvar, H. Jafarkhani | A Combined Deep-Learning and Deformable-Model Approach to Fully
Automatic Segmentation of the Left Ventricle in Cardiac MRI | to appear in Medical Image Analysis | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Segmentation of the left ventricle (LV) from cardiac magnetic resonance
imaging (MRI) datasets is an essential step for calculation of clinical indices
such as ventricular volume and ejection fraction. In this work, we employ deep
learning algorithms combined with deformable models to develop and evaluate a
fully automatic segmentation tool for the LV from short-axis cardiac MRI
datasets. The method employs deep learning algorithms to learn the segmentation
task from the ground true data. Convolutional networks are employed to
automatically detect the LV chamber in MRI dataset. Stacked autoencoders are
utilized to infer the shape of the LV. The inferred shape is incorporated into
deformable models to improve the accuracy and robustness of the segmentation.
We validated our method using 45 cardiac MR datasets taken from the MICCAI 2009
LV segmentation challenge and showed that it outperforms the state-of-the art
methods. Excellent agreement with the ground truth was achieved. Validation
metrics, percentage of good contours, Dice metric, average perpendicular
distance and conformity, were computed as 96.69%, 0.94, 1.81mm and 0.86, versus
those of 79.2%-95.62%, 0.87-0.9, 1.76-2.97mm and 0.67-0.78, obtained by other
methods, respectively.
| [
{
"version": "v1",
"created": "Fri, 25 Dec 2015 03:35:15 GMT"
}
] | 2015-12-29T00:00:00 | [
[
"Avendi",
"M. R.",
""
],
[
"Kheradvar",
"A.",
""
],
[
"Jafarkhani",
"H.",
""
]
] | TITLE: A Combined Deep-Learning and Deformable-Model Approach to Fully
Automatic Segmentation of the Left Ventricle in Cardiac MRI
ABSTRACT: Segmentation of the left ventricle (LV) from cardiac magnetic resonance
imaging (MRI) datasets is an essential step for calculation of clinical indices
such as ventricular volume and ejection fraction. In this work, we employ deep
learning algorithms combined with deformable models to develop and evaluate a
fully automatic segmentation tool for the LV from short-axis cardiac MRI
datasets. The method employs deep learning algorithms to learn the segmentation
task from the ground true data. Convolutional networks are employed to
automatically detect the LV chamber in MRI dataset. Stacked autoencoders are
utilized to infer the shape of the LV. The inferred shape is incorporated into
deformable models to improve the accuracy and robustness of the segmentation.
We validated our method using 45 cardiac MR datasets taken from the MICCAI 2009
LV segmentation challenge and showed that it outperforms the state-of-the art
methods. Excellent agreement with the ground truth was achieved. Validation
metrics, percentage of good contours, Dice metric, average perpendicular
distance and conformity, were computed as 96.69%, 0.94, 1.81mm and 0.86, versus
those of 79.2%-95.62%, 0.87-0.9, 1.76-2.97mm and 0.67-0.78, obtained by other
methods, respectively.
| no_new_dataset | 0.94887 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.