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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1612.01689 | Ra\'ul D\'iaz | Ra\'ul D\'iaz, Charless C. Fowlkes | Cluster-Wise Ratio Tests for Fast Camera Localization | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Feature point matching for camera localization suffers from scalability
problems. Even when feature descriptors associated with 3D scene points are
locally unique, as coverage grows, similar or repeated features become
increasingly common. As a result, the standard distance ratio-test used to
identify reliable image feature points is overly restrictive and rejects many
good candidate matches. We propose a simple coarse-to-fine strategy that uses
conservative approximations to robust local ratio-tests that can be computed
efficiently using global approximate k-nearest neighbor search. We treat these
forward matches as votes in camera pose space and use them to prioritize
back-matching within candidate camera pose clusters, exploiting feature
co-visibility captured by clustering the 3D model camera pose graph. This
approach achieves state-of-the-art camera localization results on a variety of
popular benchmarks, outperforming several methods that use more complicated
data structures and that make more restrictive assumptions on camera pose. We
also carry out diagnostic analyses on a difficult test dataset containing
globally repetitive structure that suggest our approach successfully adapts to
the challenges of large-scale image localization.
| [
{
"version": "v1",
"created": "Tue, 6 Dec 2016 07:35:24 GMT"
},
{
"version": "v2",
"created": "Sat, 20 May 2017 18:02:46 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Díaz",
"Raúl",
""
],
[
"Fowlkes",
"Charless C.",
""
]
] | TITLE: Cluster-Wise Ratio Tests for Fast Camera Localization
ABSTRACT: Feature point matching for camera localization suffers from scalability
problems. Even when feature descriptors associated with 3D scene points are
locally unique, as coverage grows, similar or repeated features become
increasingly common. As a result, the standard distance ratio-test used to
identify reliable image feature points is overly restrictive and rejects many
good candidate matches. We propose a simple coarse-to-fine strategy that uses
conservative approximations to robust local ratio-tests that can be computed
efficiently using global approximate k-nearest neighbor search. We treat these
forward matches as votes in camera pose space and use them to prioritize
back-matching within candidate camera pose clusters, exploiting feature
co-visibility captured by clustering the 3D model camera pose graph. This
approach achieves state-of-the-art camera localization results on a variety of
popular benchmarks, outperforming several methods that use more complicated
data structures and that make more restrictive assumptions on camera pose. We
also carry out diagnostic analyses on a difficult test dataset containing
globally repetitive structure that suggest our approach successfully adapts to
the challenges of large-scale image localization.
| no_new_dataset | 0.954052 |
1612.02590 | Zike Yan | Zike Yan, Xuezhi Xiang | Scene Flow Estimation: A Survey | 51 pages, 12 figures, 10 tables, 108 references | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper is the first to review the scene flow estimation field, which
analyzes and compares methods, technical challenges, evaluation methodologies
and performance of scene flow estimation. Existing algorithms are categorized
in terms of scene representation, data source, and calculation scheme, and the
pros and cons in each category are compared briefly. The datasets and
evaluation protocols are enumerated, and the performance of the most
representative methods is presented. A future vision is illustrated with few
questions arisen for discussion. This survey presents a general introduction
and analysis of scene flow estimation.
| [
{
"version": "v1",
"created": "Thu, 8 Dec 2016 10:44:03 GMT"
},
{
"version": "v2",
"created": "Mon, 12 Dec 2016 03:25:34 GMT"
},
{
"version": "v3",
"created": "Sun, 21 May 2017 09:52:02 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Yan",
"Zike",
""
],
[
"Xiang",
"Xuezhi",
""
]
] | TITLE: Scene Flow Estimation: A Survey
ABSTRACT: This paper is the first to review the scene flow estimation field, which
analyzes and compares methods, technical challenges, evaluation methodologies
and performance of scene flow estimation. Existing algorithms are categorized
in terms of scene representation, data source, and calculation scheme, and the
pros and cons in each category are compared briefly. The datasets and
evaluation protocols are enumerated, and the performance of the most
representative methods is presented. A future vision is illustrated with few
questions arisen for discussion. This survey presents a general introduction
and analysis of scene flow estimation.
| no_new_dataset | 0.949342 |
1701.09135 | Samarth Manoj Brahmbhatt | Samarth Brahmbhatt and James Hays | DeepNav: Learning to Navigate Large Cities | CVPR 2017 camera ready version | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present DeepNav, a Convolutional Neural Network (CNN) based algorithm for
navigating large cities using locally visible street-view images. The DeepNav
agent learns to reach its destination quickly by making the correct navigation
decisions at intersections. We collect a large-scale dataset of street-view
images organized in a graph where nodes are connected by roads. This dataset
contains 10 city graphs and more than 1 million street-view images. We propose
3 supervised learning approaches for the navigation task and show how A* search
in the city graph can be used to generate supervision for the learning. Our
annotation process is fully automated using publicly available mapping services
and requires no human input. We evaluate the proposed DeepNav models on 4
held-out cities for navigating to 5 different types of destinations. Our
algorithms outperform previous work that uses hand-crafted features and Support
Vector Regression (SVR)[19].
| [
{
"version": "v1",
"created": "Tue, 31 Jan 2017 17:14:24 GMT"
},
{
"version": "v2",
"created": "Sat, 20 May 2017 22:40:26 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Brahmbhatt",
"Samarth",
""
],
[
"Hays",
"James",
""
]
] | TITLE: DeepNav: Learning to Navigate Large Cities
ABSTRACT: We present DeepNav, a Convolutional Neural Network (CNN) based algorithm for
navigating large cities using locally visible street-view images. The DeepNav
agent learns to reach its destination quickly by making the correct navigation
decisions at intersections. We collect a large-scale dataset of street-view
images organized in a graph where nodes are connected by roads. This dataset
contains 10 city graphs and more than 1 million street-view images. We propose
3 supervised learning approaches for the navigation task and show how A* search
in the city graph can be used to generate supervision for the learning. Our
annotation process is fully automated using publicly available mapping services
and requires no human input. We evaluate the proposed DeepNav models on 4
held-out cities for navigating to 5 different types of destinations. Our
algorithms outperform previous work that uses hand-crafted features and Support
Vector Regression (SVR)[19].
| new_dataset | 0.95561 |
1702.08653 | Asli Celikyilmaz | Asli Celikyilmaz and Li Deng and Lihong Li and Chong Wang | Scaffolding Networks: Incremental Learning and Teaching Through
Questioning | 11 pages + Abstract + 3 figures | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a new paradigm of learning for reasoning, understanding, and
prediction, as well as the scaffolding network to implement this paradigm. The
scaffolding network embodies an incremental learning approach that is
formulated as a teacher-student network architecture to teach machines how to
understand text and do reasoning. The key to our computational scaffolding
approach is the interactions between the teacher and the student through
sequential questioning. The student observes each sentence in the text
incrementally, and it uses an attention-based neural net to discover and
register the key information in relation to its current memory. Meanwhile, the
teacher asks questions about the observed text, and the student network gets
rewarded by correctly answering these questions. The entire network is updated
continually using reinforcement learning. Our experimental results on synthetic
and real datasets show that the scaffolding network not only outperforms
state-of-the-art methods but also learns to do reasoning in a scalable way even
with little human generated input.
| [
{
"version": "v1",
"created": "Tue, 28 Feb 2017 05:43:10 GMT"
},
{
"version": "v2",
"created": "Fri, 19 May 2017 19:45:43 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Celikyilmaz",
"Asli",
""
],
[
"Deng",
"Li",
""
],
[
"Li",
"Lihong",
""
],
[
"Wang",
"Chong",
""
]
] | TITLE: Scaffolding Networks: Incremental Learning and Teaching Through
Questioning
ABSTRACT: We introduce a new paradigm of learning for reasoning, understanding, and
prediction, as well as the scaffolding network to implement this paradigm. The
scaffolding network embodies an incremental learning approach that is
formulated as a teacher-student network architecture to teach machines how to
understand text and do reasoning. The key to our computational scaffolding
approach is the interactions between the teacher and the student through
sequential questioning. The student observes each sentence in the text
incrementally, and it uses an attention-based neural net to discover and
register the key information in relation to its current memory. Meanwhile, the
teacher asks questions about the observed text, and the student network gets
rewarded by correctly answering these questions. The entire network is updated
continually using reinforcement learning. Our experimental results on synthetic
and real datasets show that the scaffolding network not only outperforms
state-of-the-art methods but also learns to do reasoning in a scalable way even
with little human generated input.
| no_new_dataset | 0.949389 |
1703.01789 | Jongpil Lee | Jongpil Lee, Jiyoung Park, Keunhyoung Luke Kim, Juhan Nam | Sample-level Deep Convolutional Neural Networks for Music Auto-tagging
Using Raw Waveforms | 7 pages, Sound and Music Computing Conference (SMC), 2017 | null | null | null | cs.SD cs.LG cs.MM cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, the end-to-end approach that learns hierarchical representations
from raw data using deep convolutional neural networks has been successfully
explored in the image, text and speech domains. This approach was applied to
musical signals as well but has been not fully explored yet. To this end, we
propose sample-level deep convolutional neural networks which learn
representations from very small grains of waveforms (e.g. 2 or 3 samples)
beyond typical frame-level input representations. Our experiments show how deep
architectures with sample-level filters improve the accuracy in music
auto-tagging and they provide results comparable to previous state-of-the-art
performances for the Magnatagatune dataset and Million Song Dataset. In
addition, we visualize filters learned in a sample-level DCNN in each layer to
identify hierarchically learned features and show that they are sensitive to
log-scaled frequency along layer, such as mel-frequency spectrogram that is
widely used in music classification systems.
| [
{
"version": "v1",
"created": "Mon, 6 Mar 2017 09:49:48 GMT"
},
{
"version": "v2",
"created": "Mon, 22 May 2017 04:46:36 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Lee",
"Jongpil",
""
],
[
"Park",
"Jiyoung",
""
],
[
"Kim",
"Keunhyoung Luke",
""
],
[
"Nam",
"Juhan",
""
]
] | TITLE: Sample-level Deep Convolutional Neural Networks for Music Auto-tagging
Using Raw Waveforms
ABSTRACT: Recently, the end-to-end approach that learns hierarchical representations
from raw data using deep convolutional neural networks has been successfully
explored in the image, text and speech domains. This approach was applied to
musical signals as well but has been not fully explored yet. To this end, we
propose sample-level deep convolutional neural networks which learn
representations from very small grains of waveforms (e.g. 2 or 3 samples)
beyond typical frame-level input representations. Our experiments show how deep
architectures with sample-level filters improve the accuracy in music
auto-tagging and they provide results comparable to previous state-of-the-art
performances for the Magnatagatune dataset and Million Song Dataset. In
addition, we visualize filters learned in a sample-level DCNN in each layer to
identify hierarchically learned features and show that they are sensitive to
log-scaled frequency along layer, such as mel-frequency spectrogram that is
widely used in music classification systems.
| no_new_dataset | 0.949669 |
1703.03329 | Limin Wang | Limin Wang, Yuanjun Xiong, Dahua Lin, Luc Van Gool | UntrimmedNets for Weakly Supervised Action Recognition and Detection | camera-ready version to appear in CVPR2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current action recognition methods heavily rely on trimmed videos for model
training. However, it is expensive and time-consuming to acquire a large-scale
trimmed video dataset. This paper presents a new weakly supervised
architecture, called UntrimmedNet, which is able to directly learn action
recognition models from untrimmed videos without the requirement of temporal
annotations of action instances. Our UntrimmedNet couples two important
components, the classification module and the selection module, to learn the
action models and reason about the temporal duration of action instances,
respectively. These two components are implemented with feed-forward networks,
and UntrimmedNet is therefore an end-to-end trainable architecture. We exploit
the learned models for action recognition (WSR) and detection (WSD) on the
untrimmed video datasets of THUMOS14 and ActivityNet. Although our UntrimmedNet
only employs weak supervision, our method achieves performance superior or
comparable to that of those strongly supervised approaches on these two
datasets.
| [
{
"version": "v1",
"created": "Thu, 9 Mar 2017 16:29:39 GMT"
},
{
"version": "v2",
"created": "Mon, 22 May 2017 12:38:02 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Wang",
"Limin",
""
],
[
"Xiong",
"Yuanjun",
""
],
[
"Lin",
"Dahua",
""
],
[
"Van Gool",
"Luc",
""
]
] | TITLE: UntrimmedNets for Weakly Supervised Action Recognition and Detection
ABSTRACT: Current action recognition methods heavily rely on trimmed videos for model
training. However, it is expensive and time-consuming to acquire a large-scale
trimmed video dataset. This paper presents a new weakly supervised
architecture, called UntrimmedNet, which is able to directly learn action
recognition models from untrimmed videos without the requirement of temporal
annotations of action instances. Our UntrimmedNet couples two important
components, the classification module and the selection module, to learn the
action models and reason about the temporal duration of action instances,
respectively. These two components are implemented with feed-forward networks,
and UntrimmedNet is therefore an end-to-end trainable architecture. We exploit
the learned models for action recognition (WSR) and detection (WSD) on the
untrimmed video datasets of THUMOS14 and ActivityNet. Although our UntrimmedNet
only employs weak supervision, our method achieves performance superior or
comparable to that of those strongly supervised approaches on these two
datasets.
| no_new_dataset | 0.94699 |
1704.00135 | Vadim Markovtsev | Vadim Markovtsev and Eiso Kant | Topic modeling of public repositories at scale using names in source
code | 11 pages | null | null | null | cs.PL cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Programming languages themselves have a limited number of reserved keywords
and character based tokens that define the language specification. However,
programmers have a rich use of natural language within their code through
comments, text literals and naming entities. The programmer defined names that
can be found in source code are a rich source of information to build a high
level understanding of the project. The goal of this paper is to apply topic
modeling to names used in over 13.6 million repositories and perceive the
inferred topics. One of the problems in such a study is the occurrence of
duplicate repositories not officially marked as forks (obscure forks). We show
how to address it using the same identifiers which are extracted for topic
modeling.
We open with a discussion on naming in source code, we then elaborate on our
approach to remove exact duplicate and fuzzy duplicate repositories using
Locality Sensitive Hashing on the bag-of-words model and then discuss our work
on topic modeling; and finally present the results from our data analysis
together with open-access to the source code, tools and datasets.
| [
{
"version": "v1",
"created": "Sat, 1 Apr 2017 08:16:20 GMT"
},
{
"version": "v2",
"created": "Sat, 20 May 2017 08:29:00 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Markovtsev",
"Vadim",
""
],
[
"Kant",
"Eiso",
""
]
] | TITLE: Topic modeling of public repositories at scale using names in source
code
ABSTRACT: Programming languages themselves have a limited number of reserved keywords
and character based tokens that define the language specification. However,
programmers have a rich use of natural language within their code through
comments, text literals and naming entities. The programmer defined names that
can be found in source code are a rich source of information to build a high
level understanding of the project. The goal of this paper is to apply topic
modeling to names used in over 13.6 million repositories and perceive the
inferred topics. One of the problems in such a study is the occurrence of
duplicate repositories not officially marked as forks (obscure forks). We show
how to address it using the same identifiers which are extracted for topic
modeling.
We open with a discussion on naming in source code, we then elaborate on our
approach to remove exact duplicate and fuzzy duplicate repositories using
Locality Sensitive Hashing on the bag-of-words model and then discuss our work
on topic modeling; and finally present the results from our data analysis
together with open-access to the source code, tools and datasets.
| no_new_dataset | 0.948728 |
1704.01700 | Anirban Roychowdhury | Anirban Roychowdhury | Accelerated Stochastic Quasi-Newton Optimization on Riemann Manifolds | null | null | null | null | math.OC cs.LG math.DG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose an L-BFGS optimization algorithm on Riemannian manifolds using
minibatched stochastic variance reduction techniques for fast convergence with
constant step sizes, without resorting to linesearch methods designed to
satisfy Wolfe conditions. We provide a new convergence proof for strongly
convex functions without using curvature conditions on the manifold, as well as
a convergence discussion for nonconvex functions. We discuss a couple of ways
to obtain the correction pairs used to calculate the product of the gradient
with the inverse Hessian, and empirically demonstrate their use in synthetic
experiments on computation of Karcher means for symmetric positive definite
matrices and leading eigenvalues of large scale data matrices. We compare our
method to VR-PCA for the latter experiment, along with Riemannian SVRG for both
cases, and show strong convergence results for a range of datasets.
| [
{
"version": "v1",
"created": "Thu, 6 Apr 2017 03:34:29 GMT"
},
{
"version": "v2",
"created": "Wed, 12 Apr 2017 22:02:30 GMT"
},
{
"version": "v3",
"created": "Mon, 22 May 2017 15:02:02 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Roychowdhury",
"Anirban",
""
]
] | TITLE: Accelerated Stochastic Quasi-Newton Optimization on Riemann Manifolds
ABSTRACT: We propose an L-BFGS optimization algorithm on Riemannian manifolds using
minibatched stochastic variance reduction techniques for fast convergence with
constant step sizes, without resorting to linesearch methods designed to
satisfy Wolfe conditions. We provide a new convergence proof for strongly
convex functions without using curvature conditions on the manifold, as well as
a convergence discussion for nonconvex functions. We discuss a couple of ways
to obtain the correction pairs used to calculate the product of the gradient
with the inverse Hessian, and empirically demonstrate their use in synthetic
experiments on computation of Karcher means for symmetric positive definite
matrices and leading eigenvalues of large scale data matrices. We compare our
method to VR-PCA for the latter experiment, along with Riemannian SVRG for both
cases, and show strong convergence results for a range of datasets.
| no_new_dataset | 0.945951 |
1704.02703 | Lei Bi | Lei Bi, Jinman Kim, Ashnil Kumar, Dagan Feng | Automatic Liver Lesion Detection using Cascaded Deep Residual Networks | Submission for 2017 ISBI LiTS Challenge | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic segmentation of liver lesions is a fundamental requirement towards
the creation of computer aided diagnosis (CAD) and decision support systems
(CDS). Traditional segmentation approaches depend heavily upon hand-crafted
features and a priori knowledge of the user. As such, these methods are
difficult to adopt within a clinical environment. Recently, deep learning
methods based on fully convolutional networks (FCNs) have been successful in
many segmentation problems primarily because they leverage a large labelled
dataset to hierarchically learn the features that best correspond to the
shallow visual appearance as well as the deep semantics of the areas to be
segmented. However, FCNs based on a 16 layer VGGNet architecture have limited
capacity to add additional layers. Therefore, it is challenging to learn more
discriminative features among different classes for FCNs. In this study, we
overcome these limitations using deep residual networks (ResNet) to segment
liver lesions. ResNet contain skip connections between convolutional layers,
which solved the problem of the training degradation of training accuracy in
very deep networks and thereby enables the use of additional layers for
learning more discriminative features. In addition, we achieve more precise
boundary definitions through a novel cascaded ResNet architecture with
multi-scale fusion to gradually learn and infer the boundaries of both the
liver and the liver lesions. Our proposed method achieved 4th place in the ISBI
2017 Liver Tumor Segmentation Challenge by the submission deadline.
| [
{
"version": "v1",
"created": "Mon, 10 Apr 2017 04:05:50 GMT"
},
{
"version": "v2",
"created": "Sun, 21 May 2017 02:58:40 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Bi",
"Lei",
""
],
[
"Kim",
"Jinman",
""
],
[
"Kumar",
"Ashnil",
""
],
[
"Feng",
"Dagan",
""
]
] | TITLE: Automatic Liver Lesion Detection using Cascaded Deep Residual Networks
ABSTRACT: Automatic segmentation of liver lesions is a fundamental requirement towards
the creation of computer aided diagnosis (CAD) and decision support systems
(CDS). Traditional segmentation approaches depend heavily upon hand-crafted
features and a priori knowledge of the user. As such, these methods are
difficult to adopt within a clinical environment. Recently, deep learning
methods based on fully convolutional networks (FCNs) have been successful in
many segmentation problems primarily because they leverage a large labelled
dataset to hierarchically learn the features that best correspond to the
shallow visual appearance as well as the deep semantics of the areas to be
segmented. However, FCNs based on a 16 layer VGGNet architecture have limited
capacity to add additional layers. Therefore, it is challenging to learn more
discriminative features among different classes for FCNs. In this study, we
overcome these limitations using deep residual networks (ResNet) to segment
liver lesions. ResNet contain skip connections between convolutional layers,
which solved the problem of the training degradation of training accuracy in
very deep networks and thereby enables the use of additional layers for
learning more discriminative features. In addition, we achieve more precise
boundary definitions through a novel cascaded ResNet architecture with
multi-scale fusion to gradually learn and infer the boundaries of both the
liver and the liver lesions. Our proposed method achieved 4th place in the ISBI
2017 Liver Tumor Segmentation Challenge by the submission deadline.
| no_new_dataset | 0.946101 |
1704.04599 | Arkan Al-Hamodi | Arkan A. G. Al-Hamodi, Songfeng Lu | A novel approach for fast mining frequent itemsets use N-list structure
based on MapReduce | 11 pages, 10 figures | null | null | null | cs.DC cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Frequent Pattern Mining is a one field of the most significant topics in data
mining. In recent years, many algorithms have been proposed for mining frequent
itemsets. A new algorithm has been presented for mining frequent itemsets based
on N-list data structure called Prepost algorithm. The Prepost algorithm is
enhanced by implementing compact PPC-tree with the general tree. Prepost
algorithm can only find a frequent itemsets with required (pre-order and
post-order) for each node. In this chapter, we improved prepost algorithm based
on Hadoop platform (HPrepost), proposed using the Mapreduce programming model.
The main goals of proposed method are efficient mining frequent itemsets
requiring less running time and memory usage. We have conduct experiments for
the proposed scheme to compare with another algorithms. With dense datasets,
which have a large average length of transactions, HPrepost is more effective
than frequent itemsets algorithms in terms of execution time and memory usage
for all min-sup. Generally, our algorithm outperforms algorithms in terms of
runtime and memory usage with small thresholds and large datasets.
| [
{
"version": "v1",
"created": "Sat, 15 Apr 2017 07:23:40 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Al-Hamodi",
"Arkan A. G.",
""
],
[
"Lu",
"Songfeng",
""
]
] | TITLE: A novel approach for fast mining frequent itemsets use N-list structure
based on MapReduce
ABSTRACT: Frequent Pattern Mining is a one field of the most significant topics in data
mining. In recent years, many algorithms have been proposed for mining frequent
itemsets. A new algorithm has been presented for mining frequent itemsets based
on N-list data structure called Prepost algorithm. The Prepost algorithm is
enhanced by implementing compact PPC-tree with the general tree. Prepost
algorithm can only find a frequent itemsets with required (pre-order and
post-order) for each node. In this chapter, we improved prepost algorithm based
on Hadoop platform (HPrepost), proposed using the Mapreduce programming model.
The main goals of proposed method are efficient mining frequent itemsets
requiring less running time and memory usage. We have conduct experiments for
the proposed scheme to compare with another algorithms. With dense datasets,
which have a large average length of transactions, HPrepost is more effective
than frequent itemsets algorithms in terms of execution time and memory usage
for all min-sup. Generally, our algorithm outperforms algorithms in terms of
runtime and memory usage with small thresholds and large datasets.
| no_new_dataset | 0.948058 |
1705.05183 | Sahil Manchanda | Sahil Manchanda and Ashish Anand | Representation learning of drug and disease terms for drug repositioning | Accepted to appear in 3rd IEEE International Conference on
Cybernetics (Spl Session: Deep Learning for Prediction and Estimation) | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Drug repositioning (DR) refers to identification of novel indications for the
approved drugs. The requirement of huge investment of time as well as money and
risk of failure in clinical trials have led to surge in interest in drug
repositioning. DR exploits two major aspects associated with drugs and
diseases: existence of similarity among drugs and among diseases due to their
shared involved genes or pathways or common biological effects. Existing
methods of identifying drug-disease association majorly rely on the information
available in the structured databases only. On the other hand, abundant
information available in form of free texts in biomedical research articles are
not being fully exploited. Word-embedding or obtaining vector representation of
words from a large corpora of free texts using neural network methods have been
shown to give significant performance for several natural language processing
tasks. In this work we propose a novel way of representation learning to obtain
features of drugs and diseases by combining complementary information available
in unstructured texts and structured datasets. Next we use matrix completion
approach on these feature vectors to learn projection matrix between drug and
disease vector spaces. The proposed method has shown competitive performance
with state-of-the-art methods. Further, the case studies on Alzheimer's and
Hypertension diseases have shown that the predicted associations are matching
with the existing knowledge.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 12:29:52 GMT"
},
{
"version": "v2",
"created": "Sat, 20 May 2017 12:29:56 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Manchanda",
"Sahil",
""
],
[
"Anand",
"Ashish",
""
]
] | TITLE: Representation learning of drug and disease terms for drug repositioning
ABSTRACT: Drug repositioning (DR) refers to identification of novel indications for the
approved drugs. The requirement of huge investment of time as well as money and
risk of failure in clinical trials have led to surge in interest in drug
repositioning. DR exploits two major aspects associated with drugs and
diseases: existence of similarity among drugs and among diseases due to their
shared involved genes or pathways or common biological effects. Existing
methods of identifying drug-disease association majorly rely on the information
available in the structured databases only. On the other hand, abundant
information available in form of free texts in biomedical research articles are
not being fully exploited. Word-embedding or obtaining vector representation of
words from a large corpora of free texts using neural network methods have been
shown to give significant performance for several natural language processing
tasks. In this work we propose a novel way of representation learning to obtain
features of drugs and diseases by combining complementary information available
in unstructured texts and structured datasets. Next we use matrix completion
approach on these feature vectors to learn projection matrix between drug and
disease vector spaces. The proposed method has shown competitive performance
with state-of-the-art methods. Further, the case studies on Alzheimer's and
Hypertension diseases have shown that the predicted associations are matching
with the existing knowledge.
| no_new_dataset | 0.9462 |
1705.07202 | Lei Cai | Lei Cai and Hongyang Gao and Shuiwang Ji | Multi-Stage Variational Auto-Encoders for Coarse-to-Fine Image
Generation | null | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Variational auto-encoder (VAE) is a powerful unsupervised learning framework
for image generation. One drawback of VAE is that it generates blurry images
due to its Gaussianity assumption and thus L2 loss. To allow the generation of
high quality images by VAE, we increase the capacity of decoder network by
employing residual blocks and skip connections, which also enable efficient
optimization. To overcome the limitation of L2 loss, we propose to generate
images in a multi-stage manner from coarse to fine. In the simplest case, the
proposed multi-stage VAE divides the decoder into two components in which the
second component generates refined images based on the course images generated
by the first component. Since the second component is independent of the VAE
model, it can employ other loss functions beyond the L2 loss and different
model architectures. The proposed framework can be easily generalized to
contain more than two components. Experiment results on the MNIST and CelebA
datasets demonstrate that the proposed multi-stage VAE can generate sharper
images as compared to those from the original VAE.
| [
{
"version": "v1",
"created": "Fri, 19 May 2017 21:51:30 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Cai",
"Lei",
""
],
[
"Gao",
"Hongyang",
""
],
[
"Ji",
"Shuiwang",
""
]
] | TITLE: Multi-Stage Variational Auto-Encoders for Coarse-to-Fine Image
Generation
ABSTRACT: Variational auto-encoder (VAE) is a powerful unsupervised learning framework
for image generation. One drawback of VAE is that it generates blurry images
due to its Gaussianity assumption and thus L2 loss. To allow the generation of
high quality images by VAE, we increase the capacity of decoder network by
employing residual blocks and skip connections, which also enable efficient
optimization. To overcome the limitation of L2 loss, we propose to generate
images in a multi-stage manner from coarse to fine. In the simplest case, the
proposed multi-stage VAE divides the decoder into two components in which the
second component generates refined images based on the course images generated
by the first component. Since the second component is independent of the VAE
model, it can employ other loss functions beyond the L2 loss and different
model architectures. The proposed framework can be easily generalized to
contain more than two components. Experiment results on the MNIST and CelebA
datasets demonstrate that the proposed multi-stage VAE can generate sharper
images as compared to those from the original VAE.
| no_new_dataset | 0.94868 |
1705.07256 | Samet Oymak | Samet Oymak, Mehrdad Mahdavi, Jiasi Chen | Learning Feature Nonlinearities with Non-Convex Regularized Binned
Regression | 22 pages, 7 figures | null | null | null | cs.LG cs.IT math.IT math.OC stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For various applications, the relations between the dependent and independent
variables are highly nonlinear. Consequently, for large scale complex problems,
neural networks and regression trees are commonly preferred over linear models
such as Lasso. This work proposes learning the feature nonlinearities by
binning feature values and finding the best fit in each quantile using
non-convex regularized linear regression. The algorithm first captures the
dependence between neighboring quantiles by enforcing smoothness via
piecewise-constant/linear approximation and then selects a sparse subset of
good features. We prove that the proposed algorithm is statistically and
computationally efficient. In particular, it achieves linear rate of
convergence while requiring near-minimal number of samples. Evaluations on
synthetic and real datasets demonstrate that algorithm is competitive with
current state-of-the-art and accurately learns feature nonlinearities. Finally,
we explore an interesting connection between the binning stage of our algorithm
and sparse Johnson-Lindenstrauss matrices.
| [
{
"version": "v1",
"created": "Sat, 20 May 2017 03:46:32 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Oymak",
"Samet",
""
],
[
"Mahdavi",
"Mehrdad",
""
],
[
"Chen",
"Jiasi",
""
]
] | TITLE: Learning Feature Nonlinearities with Non-Convex Regularized Binned
Regression
ABSTRACT: For various applications, the relations between the dependent and independent
variables are highly nonlinear. Consequently, for large scale complex problems,
neural networks and regression trees are commonly preferred over linear models
such as Lasso. This work proposes learning the feature nonlinearities by
binning feature values and finding the best fit in each quantile using
non-convex regularized linear regression. The algorithm first captures the
dependence between neighboring quantiles by enforcing smoothness via
piecewise-constant/linear approximation and then selects a sparse subset of
good features. We prove that the proposed algorithm is statistically and
computationally efficient. In particular, it achieves linear rate of
convergence while requiring near-minimal number of samples. Evaluations on
synthetic and real datasets demonstrate that algorithm is competitive with
current state-of-the-art and accurately learns feature nonlinearities. Finally,
we explore an interesting connection between the binning stage of our algorithm
and sparse Johnson-Lindenstrauss matrices.
| no_new_dataset | 0.947527 |
1705.07258 | Ziqi Yan | Ziqi Yan, Jiqiang Liu, Gang Li, Zhen Han, Shuo Qiu | PrivMin: Differentially Private MinHash for Jaccard Similarity
Computation | 27 pages, 6 figures, 4 tables | null | null | null | cs.DS cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In many industrial applications of big data, the Jaccard Similarity
Computation has been widely used to measure the distance between two profiles
or sets respectively owned by two users. Yet, one semi-honest user with
unpredictable knowledge may also deduce the private or sensitive information
(e.g., the existence of a single element in the original sets) of the other
user via the shared similarity. In this paper, we aim at solving the privacy
issues in Jaccard similarity computation with strict differential privacy
guarantees. To achieve this, we first define the Conditional $\epsilon$-DPSO, a
relaxed differential privacy definition regarding set operations, and prove
that the MinHash-based Jaccard Similarity Computation (MH-JSC) satisfies this
definition. Then for achieving strict differential privacy in MH-JSC, we
propose the PrivMin algorithm, which consists of two private operations: 1) the
Private MinHash Value Generation that works by introducing the Exponential
noise to the generation of MinHash signature. 2) the Randomized MinHashing
Steps Selection that works by adopting Randomized Response technique to
privately select several steps within the MinHashing phase that are deployed
with the Exponential mechanism. Experiments on real datasets demonstrate that
the proposed PrivMin algorithm can successfully retain the utility of the
computed similarity while preserving privacy.
| [
{
"version": "v1",
"created": "Sat, 20 May 2017 04:09:12 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Yan",
"Ziqi",
""
],
[
"Liu",
"Jiqiang",
""
],
[
"Li",
"Gang",
""
],
[
"Han",
"Zhen",
""
],
[
"Qiu",
"Shuo",
""
]
] | TITLE: PrivMin: Differentially Private MinHash for Jaccard Similarity
Computation
ABSTRACT: In many industrial applications of big data, the Jaccard Similarity
Computation has been widely used to measure the distance between two profiles
or sets respectively owned by two users. Yet, one semi-honest user with
unpredictable knowledge may also deduce the private or sensitive information
(e.g., the existence of a single element in the original sets) of the other
user via the shared similarity. In this paper, we aim at solving the privacy
issues in Jaccard similarity computation with strict differential privacy
guarantees. To achieve this, we first define the Conditional $\epsilon$-DPSO, a
relaxed differential privacy definition regarding set operations, and prove
that the MinHash-based Jaccard Similarity Computation (MH-JSC) satisfies this
definition. Then for achieving strict differential privacy in MH-JSC, we
propose the PrivMin algorithm, which consists of two private operations: 1) the
Private MinHash Value Generation that works by introducing the Exponential
noise to the generation of MinHash signature. 2) the Randomized MinHashing
Steps Selection that works by adopting Randomized Response technique to
privately select several steps within the MinHashing phase that are deployed
with the Exponential mechanism. Experiments on real datasets demonstrate that
the proposed PrivMin algorithm can successfully retain the utility of the
computed similarity while preserving privacy.
| no_new_dataset | 0.945851 |
1705.07290 | Chandra Sekhar Seelamantula | Debabrata Mahapatra, Subhadip Mukherjee, and Chandra Sekhar
Seelamantula | Deep Sparse Coding Using Optimized Linear Expansion of Thresholds | Submission date: November 11, 2016. 19 pages; 9 figures | null | null | IEEE Transactions on Pattern Analysis and Machine Intelligence
Manuscript ID: TPAMI-2016-11-0861; | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We address the problem of reconstructing sparse signals from noisy and
compressive measurements using a feed-forward deep neural network (DNN) with an
architecture motivated by the iterative shrinkage-thresholding algorithm
(ISTA). We maintain the weights and biases of the network links as prescribed
by ISTA and model the nonlinear activation function using a linear expansion of
thresholds (LET), which has been very successful in image denoising and
deconvolution. The optimal set of coefficients of the parametrized activation
is learned over a training dataset containing measurement-sparse signal pairs,
corresponding to a fixed sensing matrix. For training, we develop an efficient
second-order algorithm, which requires only matrix-vector product computations
in every training epoch (Hessian-free optimization) and offers superior
convergence performance than gradient-descent optimization. Subsequently, we
derive an improved network architecture inspired by FISTA, a faster version of
ISTA, to achieve similar signal estimation performance with about 50% of the
number of layers. The resulting architecture turns out to be a deep residual
network, which has recently been shown to exhibit superior performance in
several visual recognition tasks. Numerical experiments demonstrate that the
proposed DNN architectures lead to 3 to 4 dB improvement in the reconstruction
signal-to-noise ratio (SNR), compared with the state-of-the-art sparse coding
algorithms.
| [
{
"version": "v1",
"created": "Sat, 20 May 2017 11:14:39 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Mahapatra",
"Debabrata",
""
],
[
"Mukherjee",
"Subhadip",
""
],
[
"Seelamantula",
"Chandra Sekhar",
""
]
] | TITLE: Deep Sparse Coding Using Optimized Linear Expansion of Thresholds
ABSTRACT: We address the problem of reconstructing sparse signals from noisy and
compressive measurements using a feed-forward deep neural network (DNN) with an
architecture motivated by the iterative shrinkage-thresholding algorithm
(ISTA). We maintain the weights and biases of the network links as prescribed
by ISTA and model the nonlinear activation function using a linear expansion of
thresholds (LET), which has been very successful in image denoising and
deconvolution. The optimal set of coefficients of the parametrized activation
is learned over a training dataset containing measurement-sparse signal pairs,
corresponding to a fixed sensing matrix. For training, we develop an efficient
second-order algorithm, which requires only matrix-vector product computations
in every training epoch (Hessian-free optimization) and offers superior
convergence performance than gradient-descent optimization. Subsequently, we
derive an improved network architecture inspired by FISTA, a faster version of
ISTA, to achieve similar signal estimation performance with about 50% of the
number of layers. The resulting architecture turns out to be a deep residual
network, which has recently been shown to exhibit superior performance in
several visual recognition tasks. Numerical experiments demonstrate that the
proposed DNN architectures lead to 3 to 4 dB improvement in the reconstruction
signal-to-noise ratio (SNR), compared with the state-of-the-art sparse coding
algorithms.
| no_new_dataset | 0.945096 |
1705.07311 | Mohammad Aliannejadi | Mohammad Aliannejadi, Ida Mele, and Fabio Crestani | Personalized Ranking for Context-Aware Venue Suggestion | The 32nd ACM SIGAPP Symposium On Applied Computing (SAC), Marrakech,
Morocco, April 4-6, 2017 | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Making personalized and context-aware suggestions of venues to the users is
very crucial in venue recommendation. These suggestions are often based on
matching the venues' features with the users' preferences, which can be
collected from previously visited locations. In this paper we present a novel
user-modeling approach which relies on a set of scoring functions for making
personalized suggestions of venues based on venues content and reviews as well
as users context. Our experiments, conducted on the dataset of the TREC
Contextual Suggestion Track, prove that our methodology outperforms
state-of-the-art approaches by a significant margin.
| [
{
"version": "v1",
"created": "Sat, 20 May 2017 14:21:02 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Aliannejadi",
"Mohammad",
""
],
[
"Mele",
"Ida",
""
],
[
"Crestani",
"Fabio",
""
]
] | TITLE: Personalized Ranking for Context-Aware Venue Suggestion
ABSTRACT: Making personalized and context-aware suggestions of venues to the users is
very crucial in venue recommendation. These suggestions are often based on
matching the venues' features with the users' preferences, which can be
collected from previously visited locations. In this paper we present a novel
user-modeling approach which relies on a set of scoring functions for making
personalized suggestions of venues based on venues content and reviews as well
as users context. Our experiments, conducted on the dataset of the TREC
Contextual Suggestion Track, prove that our methodology outperforms
state-of-the-art approaches by a significant margin.
| no_new_dataset | 0.951639 |
1705.07366 | Jeffrey Humpherys | Kevin Miller, Chris Hettinger, Jeffrey Humpherys, Tyler Jarvis, and
David Kartchner | Forward Thinking: Building Deep Random Forests | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The success of deep neural networks has inspired many to wonder whether other
learners could benefit from deep, layered architectures. We present a general
framework called forward thinking for deep learning that generalizes the
architectural flexibility and sophistication of deep neural networks while also
allowing for (i) different types of learning functions in the network, other
than neurons, and (ii) the ability to adaptively deepen the network as needed
to improve results. This is done by training one layer at a time, and once a
layer is trained, the input data are mapped forward through the layer to create
a new learning problem. The process is then repeated, transforming the data
through multiple layers, one at a time, rendering a new dataset, which is
expected to be better behaved, and on which a final output layer can achieve
good performance. In the case where the neurons of deep neural nets are
replaced with decision trees, we call the result a Forward Thinking Deep Random
Forest (FTDRF). We demonstrate a proof of concept by applying FTDRF on the
MNIST dataset. We also provide a general mathematical formulation that allows
for other types of deep learning problems to be considered.
| [
{
"version": "v1",
"created": "Sat, 20 May 2017 22:39:51 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Miller",
"Kevin",
""
],
[
"Hettinger",
"Chris",
""
],
[
"Humpherys",
"Jeffrey",
""
],
[
"Jarvis",
"Tyler",
""
],
[
"Kartchner",
"David",
""
]
] | TITLE: Forward Thinking: Building Deep Random Forests
ABSTRACT: The success of deep neural networks has inspired many to wonder whether other
learners could benefit from deep, layered architectures. We present a general
framework called forward thinking for deep learning that generalizes the
architectural flexibility and sophistication of deep neural networks while also
allowing for (i) different types of learning functions in the network, other
than neurons, and (ii) the ability to adaptively deepen the network as needed
to improve results. This is done by training one layer at a time, and once a
layer is trained, the input data are mapped forward through the layer to create
a new learning problem. The process is then repeated, transforming the data
through multiple layers, one at a time, rendering a new dataset, which is
expected to be better behaved, and on which a final output layer can achieve
good performance. In the case where the neurons of deep neural nets are
replaced with decision trees, we call the result a Forward Thinking Deep Random
Forest (FTDRF). We demonstrate a proof of concept by applying FTDRF on the
MNIST dataset. We also provide a general mathematical formulation that allows
for other types of deep learning problems to be considered.
| new_dataset | 0.738009 |
1705.07522 | Hamid Tizhoosh | Morteza Babaie, Shivam Kalra, Aditya Sriram, Christopher Mitcheltree,
Shujin Zhu, Amin Khatami, Shahryar Rahnamayan, H.R. Tizhoosh | Classification and Retrieval of Digital Pathology Scans: A New Dataset | Accepted for presentation at Workshop for Computer Vision for
Microscopy Image Analysis (CVMI 2017) @ CVPR 2017, Honolulu, Hawaii | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we introduce a new dataset, \textbf{Kimia Path24}, for image
classification and retrieval in digital pathology. We use the whole scan images
of 24 different tissue textures to generate 1,325 test patches of size
1000$\times$1000 (0.5mm$\times$0.5mm). Training data can be generated according
to preferences of algorithm designer and can range from approximately 27,000 to
over 50,000 patches if the preset parameters are adopted. We propose a compound
patch-and-scan accuracy measurement that makes achieving high accuracies quite
challenging. In addition, we set the benchmarking line by applying LBP,
dictionary approach and convolutional neural nets (CNNs) and report their
results. The highest accuracy was 41.80\% for CNN.
| [
{
"version": "v1",
"created": "Mon, 22 May 2017 00:00:18 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Babaie",
"Morteza",
""
],
[
"Kalra",
"Shivam",
""
],
[
"Sriram",
"Aditya",
""
],
[
"Mitcheltree",
"Christopher",
""
],
[
"Zhu",
"Shujin",
""
],
[
"Khatami",
"Amin",
""
],
[
"Rahnamayan",
"Shahryar",
""
],
[
"Tizhoosh",
"H. R.",
""
]
] | TITLE: Classification and Retrieval of Digital Pathology Scans: A New Dataset
ABSTRACT: In this paper, we introduce a new dataset, \textbf{Kimia Path24}, for image
classification and retrieval in digital pathology. We use the whole scan images
of 24 different tissue textures to generate 1,325 test patches of size
1000$\times$1000 (0.5mm$\times$0.5mm). Training data can be generated according
to preferences of algorithm designer and can range from approximately 27,000 to
over 50,000 patches if the preset parameters are adopted. We propose a compound
patch-and-scan accuracy measurement that makes achieving high accuracies quite
challenging. In addition, we set the benchmarking line by applying LBP,
dictionary approach and convolutional neural nets (CNNs) and report their
results. The highest accuracy was 41.80\% for CNN.
| new_dataset | 0.957675 |
1705.07563 | Yuxin Su | Yuxin Su, Irwin King, Michael Lyu | Learning to Rank Using Localized Geometric Mean Metrics | To appear in SIGIR'17 | null | 10.1145/3077136.3080828 | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many learning-to-rank (LtR) algorithms focus on query-independent model, in
which query and document do not lie in the same feature space, and the rankers
rely on the feature ensemble about query-document pair instead of the
similarity between query instance and documents. However, existing algorithms
do not consider local structures in query-document feature space, and are
fragile to irrelevant noise features. In this paper, we propose a novel
Riemannian metric learning algorithm to capture the local structures and
develop a robust LtR algorithm. First, we design a concept called \textit{ideal
candidate document} to introduce metric learning algorithm to query-independent
model. Previous metric learning algorithms aiming to find an optimal metric
space are only suitable for query-dependent model, in which query instance and
documents belong to the same feature space and the similarity is directly
computed from the metric space. Then we extend the new and extremely fast
global Geometric Mean Metric Learning (GMML) algorithm to develop a localized
GMML, namely L-GMML. Based on the combination of local learned metrics, we
employ the popular Normalized Discounted Cumulative Gain~(NDCG) scorer and
Weighted Approximate Rank Pairwise (WARP) loss to optimize the \textit{ideal
candidate document} for each query candidate set. Finally, we can quickly
evaluate all candidates via the similarity between the \textit{ideal candidate
document} and other candidates. By leveraging the ability of metric learning
algorithms to describe the complex structural information, our approach gives
us a principled and efficient way to perform LtR tasks. The experiments on
real-world datasets demonstrate that our proposed L-GMML algorithm outperforms
the state-of-the-art metric learning to rank methods and the stylish
query-independent LtR algorithms regarding accuracy and computational
efficiency.
| [
{
"version": "v1",
"created": "Mon, 22 May 2017 05:46:44 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Su",
"Yuxin",
""
],
[
"King",
"Irwin",
""
],
[
"Lyu",
"Michael",
""
]
] | TITLE: Learning to Rank Using Localized Geometric Mean Metrics
ABSTRACT: Many learning-to-rank (LtR) algorithms focus on query-independent model, in
which query and document do not lie in the same feature space, and the rankers
rely on the feature ensemble about query-document pair instead of the
similarity between query instance and documents. However, existing algorithms
do not consider local structures in query-document feature space, and are
fragile to irrelevant noise features. In this paper, we propose a novel
Riemannian metric learning algorithm to capture the local structures and
develop a robust LtR algorithm. First, we design a concept called \textit{ideal
candidate document} to introduce metric learning algorithm to query-independent
model. Previous metric learning algorithms aiming to find an optimal metric
space are only suitable for query-dependent model, in which query instance and
documents belong to the same feature space and the similarity is directly
computed from the metric space. Then we extend the new and extremely fast
global Geometric Mean Metric Learning (GMML) algorithm to develop a localized
GMML, namely L-GMML. Based on the combination of local learned metrics, we
employ the popular Normalized Discounted Cumulative Gain~(NDCG) scorer and
Weighted Approximate Rank Pairwise (WARP) loss to optimize the \textit{ideal
candidate document} for each query candidate set. Finally, we can quickly
evaluate all candidates via the similarity between the \textit{ideal candidate
document} and other candidates. By leveraging the ability of metric learning
algorithms to describe the complex structural information, our approach gives
us a principled and efficient way to perform LtR tasks. The experiments on
real-world datasets demonstrate that our proposed L-GMML algorithm outperforms
the state-of-the-art metric learning to rank methods and the stylish
query-independent LtR algorithms regarding accuracy and computational
efficiency.
| no_new_dataset | 0.949949 |
1705.07609 | Hassan Foroosh | Yuping Shen and Hassan Foroosh | View-Invariant Recognition of Action Style Self-Dissimilarity | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-similarity was recently introduced as a measure of inter-class
congruence for classification of actions. Herein, we investigate the dual
problem of intra-class dissimilarity for classification of action styles. We
introduce self-dissimilarity matrices that discriminate between same actions
performed by different subjects regardless of viewing direction and camera
parameters. We investigate two frameworks using these invariant style
dissimilarity measures based on Principal Component Analysis (PCA) and Fisher
Discriminant Analysis (FDA). Extensive experiments performed on IXMAS dataset
indicate remarkably good discriminant characteristics for the proposed
invariant measures for gender recognition from video data.
| [
{
"version": "v1",
"created": "Mon, 22 May 2017 08:38:19 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Shen",
"Yuping",
""
],
[
"Foroosh",
"Hassan",
""
]
] | TITLE: View-Invariant Recognition of Action Style Self-Dissimilarity
ABSTRACT: Self-similarity was recently introduced as a measure of inter-class
congruence for classification of actions. Herein, we investigate the dual
problem of intra-class dissimilarity for classification of action styles. We
introduce self-dissimilarity matrices that discriminate between same actions
performed by different subjects regardless of viewing direction and camera
parameters. We investigate two frameworks using these invariant style
dissimilarity measures based on Principal Component Analysis (PCA) and Fisher
Discriminant Analysis (FDA). Extensive experiments performed on IXMAS dataset
indicate remarkably good discriminant characteristics for the proposed
invariant measures for gender recognition from video data.
| no_new_dataset | 0.942981 |
1705.07692 | Yunlong Yu | Zhong Ji, Yunxin Sun, Yulong Yu, Jichang Guo, and Yanwei Pang | Semantic Softmax Loss for Zero-Shot Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A typical pipeline for Zero-Shot Learning (ZSL) is to integrate the visual
features and the class semantic descriptors into a multimodal framework with a
linear or bilinear model. However, the visual features and the class semantic
descriptors locate in different structural spaces, a linear or bilinear model
can not capture the semantic interactions between different modalities well. In
this letter, we propose a nonlinear approach to impose ZSL as a multi-class
classification problem via a Semantic Softmax Loss by embedding the class
semantic descriptors into the softmax layer of multi-class classification
network. To narrow the structural differences between the visual features and
semantic descriptors, we further use an L2 normalization constraint to the
differences between the visual features and visual prototypes reconstructed
with the semantic descriptors. The results on three benchmark datasets, i.e.,
AwA, CUB and SUN demonstrate the proposed approach can boost the performances
steadily and achieve the state-of-the-art performance for both zero-shot
classification and zero-shot retrieval.
| [
{
"version": "v1",
"created": "Mon, 22 May 2017 12:26:04 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Ji",
"Zhong",
""
],
[
"Sun",
"Yunxin",
""
],
[
"Yu",
"Yulong",
""
],
[
"Guo",
"Jichang",
""
],
[
"Pang",
"Yanwei",
""
]
] | TITLE: Semantic Softmax Loss for Zero-Shot Learning
ABSTRACT: A typical pipeline for Zero-Shot Learning (ZSL) is to integrate the visual
features and the class semantic descriptors into a multimodal framework with a
linear or bilinear model. However, the visual features and the class semantic
descriptors locate in different structural spaces, a linear or bilinear model
can not capture the semantic interactions between different modalities well. In
this letter, we propose a nonlinear approach to impose ZSL as a multi-class
classification problem via a Semantic Softmax Loss by embedding the class
semantic descriptors into the softmax layer of multi-class classification
network. To narrow the structural differences between the visual features and
semantic descriptors, we further use an L2 normalization constraint to the
differences between the visual features and visual prototypes reconstructed
with the semantic descriptors. The results on three benchmark datasets, i.e.,
AwA, CUB and SUN demonstrate the proposed approach can boost the performances
steadily and achieve the state-of-the-art performance for both zero-shot
classification and zero-shot retrieval.
| no_new_dataset | 0.945801 |
1705.07818 | Chenliang Xu | Li Ding and Chenliang Xu | TricorNet: A Hybrid Temporal Convolutional and Recurrent Network for
Video Action Segmentation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Action segmentation as a milestone towards building automatic systems to
understand untrimmed videos has received considerable attention in the recent
years. It is typically being modeled as a sequence labeling problem but
contains intrinsic and sufficient differences than text parsing or speech
processing. In this paper, we introduce a novel hybrid temporal convolutional
and recurrent network (TricorNet), which has an encoder-decoder architecture:
the encoder consists of a hierarchy of temporal convolutional kernels that
capture the local motion changes of different actions; the decoder is a
hierarchy of recurrent neural networks that are able to learn and memorize
long-term action dependencies after the encoding stage. Our model is simple but
extremely effective in terms of video sequence labeling. The experimental
results on three public action segmentation datasets have shown that the
proposed model achieves superior performance over the state of the art.
| [
{
"version": "v1",
"created": "Mon, 22 May 2017 15:55:08 GMT"
}
] | 2017-05-23T00:00:00 | [
[
"Ding",
"Li",
""
],
[
"Xu",
"Chenliang",
""
]
] | TITLE: TricorNet: A Hybrid Temporal Convolutional and Recurrent Network for
Video Action Segmentation
ABSTRACT: Action segmentation as a milestone towards building automatic systems to
understand untrimmed videos has received considerable attention in the recent
years. It is typically being modeled as a sequence labeling problem but
contains intrinsic and sufficient differences than text parsing or speech
processing. In this paper, we introduce a novel hybrid temporal convolutional
and recurrent network (TricorNet), which has an encoder-decoder architecture:
the encoder consists of a hierarchy of temporal convolutional kernels that
capture the local motion changes of different actions; the decoder is a
hierarchy of recurrent neural networks that are able to learn and memorize
long-term action dependencies after the encoding stage. Our model is simple but
extremely effective in terms of video sequence labeling. The experimental
results on three public action segmentation datasets have shown that the
proposed model achieves superior performance over the state of the art.
| no_new_dataset | 0.946597 |
1604.06414 | Da Zheng | Da Zheng, Disa Mhembere, Joshua T. Vogelstein, Carey E. Priebe, Randal
Burns | FlashR: R-Programmed Parallel and Scalable Machine Learning using SSDs | null | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | R is one of the most popular programming languages for statistics and machine
learning, but the R framework is relatively slow and unable to scale to large
datasets. The general approach for speeding up an implementation in R is to
implement the algorithms in C or FORTRAN and provide an R wrapper. FlashR takes
a different approach: it executes R code in parallel and scales the code beyond
memory capacity by utilizing solid-state drives (SSDs) automatically. It
provides a small number of generalized operations (GenOps) upon which we
reimplement a large number of matrix functions in the R base package. As such,
FlashR parallelizes and scales existing R code with little/no modification. To
reduce data movement between CPU and SSDs, FlashR evaluates matrix operations
lazily, fuses operations at runtime, and uses cache-aware, two-level matrix
partitioning. We evaluate FlashR on a variety of machine learning and
statistics algorithms on inputs of up to four billion data points. FlashR
out-of-core tracks closely the performance of FlashR in-memory. The R code for
machine learning algorithms executed in FlashR outperforms the in-memory
execution of H2O and Spark MLlib by a factor of 2-10 and outperforms Revolution
R Open by more than an order of magnitude.
| [
{
"version": "v1",
"created": "Thu, 21 Apr 2016 18:43:38 GMT"
},
{
"version": "v2",
"created": "Sat, 30 Apr 2016 00:43:50 GMT"
},
{
"version": "v3",
"created": "Wed, 18 May 2016 13:42:30 GMT"
},
{
"version": "v4",
"created": "Thu, 18 May 2017 23:28:01 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Zheng",
"Da",
""
],
[
"Mhembere",
"Disa",
""
],
[
"Vogelstein",
"Joshua T.",
""
],
[
"Priebe",
"Carey E.",
""
],
[
"Burns",
"Randal",
""
]
] | TITLE: FlashR: R-Programmed Parallel and Scalable Machine Learning using SSDs
ABSTRACT: R is one of the most popular programming languages for statistics and machine
learning, but the R framework is relatively slow and unable to scale to large
datasets. The general approach for speeding up an implementation in R is to
implement the algorithms in C or FORTRAN and provide an R wrapper. FlashR takes
a different approach: it executes R code in parallel and scales the code beyond
memory capacity by utilizing solid-state drives (SSDs) automatically. It
provides a small number of generalized operations (GenOps) upon which we
reimplement a large number of matrix functions in the R base package. As such,
FlashR parallelizes and scales existing R code with little/no modification. To
reduce data movement between CPU and SSDs, FlashR evaluates matrix operations
lazily, fuses operations at runtime, and uses cache-aware, two-level matrix
partitioning. We evaluate FlashR on a variety of machine learning and
statistics algorithms on inputs of up to four billion data points. FlashR
out-of-core tracks closely the performance of FlashR in-memory. The R code for
machine learning algorithms executed in FlashR outperforms the in-memory
execution of H2O and Spark MLlib by a factor of 2-10 and outperforms Revolution
R Open by more than an order of magnitude.
| no_new_dataset | 0.932515 |
1610.03023 | Weixun Zhou | Weixun Zhou, Shawn Newsam, Congmin Li, Zhenfeng Shao | Learning Low Dimensional Convolutional Neural Networks for
High-Resolution Remote Sensing Image Retrieval | null | Remote Sens., 9(5), 489 (2017) | 10.3390/rs9050489 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning powerful feature representations for image retrieval has always been
a challenging task in the field of remote sensing. Traditional methods focus on
extracting low-level hand-crafted features which are not only time-consuming
but also tend to achieve unsatisfactory performance due to the content
complexity of remote sensing images. In this paper, we investigate how to
extract deep feature representations based on convolutional neural networks
(CNN) for high-resolution remote sensing image retrieval (HRRSIR). To this end,
two effective schemes are proposed to generate powerful feature representations
for HRRSIR. In the first scheme, the deep features are extracted from the
fully-connected and convolutional layers of the pre-trained CNN models,
respectively; in the second scheme, we propose a novel CNN architecture based
on conventional convolution layers and a three-layer perceptron. The novel CNN
model is then trained on a large remote sensing dataset to learn low
dimensional features. The two schemes are evaluated on several public and
challenging datasets, and the results indicate that the proposed schemes and in
particular the novel CNN are able to achieve state-of-the-art performance.
| [
{
"version": "v1",
"created": "Mon, 10 Oct 2016 18:45:30 GMT"
},
{
"version": "v2",
"created": "Fri, 30 Dec 2016 19:04:58 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Zhou",
"Weixun",
""
],
[
"Newsam",
"Shawn",
""
],
[
"Li",
"Congmin",
""
],
[
"Shao",
"Zhenfeng",
""
]
] | TITLE: Learning Low Dimensional Convolutional Neural Networks for
High-Resolution Remote Sensing Image Retrieval
ABSTRACT: Learning powerful feature representations for image retrieval has always been
a challenging task in the field of remote sensing. Traditional methods focus on
extracting low-level hand-crafted features which are not only time-consuming
but also tend to achieve unsatisfactory performance due to the content
complexity of remote sensing images. In this paper, we investigate how to
extract deep feature representations based on convolutional neural networks
(CNN) for high-resolution remote sensing image retrieval (HRRSIR). To this end,
two effective schemes are proposed to generate powerful feature representations
for HRRSIR. In the first scheme, the deep features are extracted from the
fully-connected and convolutional layers of the pre-trained CNN models,
respectively; in the second scheme, we propose a novel CNN architecture based
on conventional convolution layers and a three-layer perceptron. The novel CNN
model is then trained on a large remote sensing dataset to learn low
dimensional features. The two schemes are evaluated on several public and
challenging datasets, and the results indicate that the proposed schemes and in
particular the novel CNN are able to achieve state-of-the-art performance.
| no_new_dataset | 0.955858 |
1704.04133 | Devinder Kumar | Devinder Kumar, Alexander Wong, Graham W. Taylor | Explaining the Unexplained: A CLass-Enhanced Attentive Response (CLEAR)
Approach to Understanding Deep Neural Networks | Accepted at Computer Vision and Patter Recognition Workshop (CVPR-W)
on Explainable Computer Vision, 2017 | null | null | null | cs.CV cs.AI cs.LG cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we propose CLass-Enhanced Attentive Response (CLEAR): an
approach to visualize and understand the decisions made by deep neural networks
(DNNs) given a specific input. CLEAR facilitates the visualization of attentive
regions and levels of interest of DNNs during the decision-making process. It
also enables the visualization of the most dominant classes associated with
these attentive regions of interest. As such, CLEAR can mitigate some of the
shortcomings of heatmap-based methods associated with decision ambiguity, and
allows for better insights into the decision-making process of DNNs.
Quantitative and qualitative experiments across three different datasets
demonstrate the efficacy of CLEAR for gaining a better understanding of the
inner workings of DNNs during the decision-making process.
| [
{
"version": "v1",
"created": "Thu, 13 Apr 2017 13:44:33 GMT"
},
{
"version": "v2",
"created": "Thu, 18 May 2017 18:38:06 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Kumar",
"Devinder",
""
],
[
"Wong",
"Alexander",
""
],
[
"Taylor",
"Graham W.",
""
]
] | TITLE: Explaining the Unexplained: A CLass-Enhanced Attentive Response (CLEAR)
Approach to Understanding Deep Neural Networks
ABSTRACT: In this work, we propose CLass-Enhanced Attentive Response (CLEAR): an
approach to visualize and understand the decisions made by deep neural networks
(DNNs) given a specific input. CLEAR facilitates the visualization of attentive
regions and levels of interest of DNNs during the decision-making process. It
also enables the visualization of the most dominant classes associated with
these attentive regions of interest. As such, CLEAR can mitigate some of the
shortcomings of heatmap-based methods associated with decision ambiguity, and
allows for better insights into the decision-making process of DNNs.
Quantitative and qualitative experiments across three different datasets
demonstrate the efficacy of CLEAR for gaining a better understanding of the
inner workings of DNNs during the decision-making process.
| no_new_dataset | 0.951594 |
1705.01567 | Manuel G\"unther | Manuel G\"unther, Steve Cruz, Ethan M. Rudd, Terrance E. Boult | Toward Open-Set Face Recognition | Accepted for Publication in CVPR 2017 Biometrics Workshop | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Much research has been conducted on both face identification and face
verification, with greater focus on the latter. Research on face identification
has mostly focused on using closed-set protocols, which assume that all probe
images used in evaluation contain identities of subjects that are enrolled in
the gallery. Real systems, however, where only a fraction of probe sample
identities are enrolled in the gallery, cannot make this closed-set assumption.
Instead, they must assume an open set of probe samples and be able to
reject/ignore those that correspond to unknown identities. In this paper, we
address the widespread misconception that thresholding verification-like scores
is a good way to solve the open-set face identification problem, by formulating
an open-set face identification protocol and evaluating different strategies
for assessing similarity. Our open-set identification protocol is based on the
canonical labeled faces in the wild (LFW) dataset. Additionally to the known
identities, we introduce the concepts of known unknowns (known, but
uninteresting persons) and unknown unknowns (people never seen before) to the
biometric community. We compare three algorithms for assessing similarity in a
deep feature space under an open-set protocol: thresholded verification-like
scores, linear discriminant analysis (LDA) scores, and an extreme value machine
(EVM) probabilities. Our findings suggest that thresholding EVM probabilities,
which are open-set by design, outperforms thresholding verification-like
scores.
| [
{
"version": "v1",
"created": "Wed, 3 May 2017 18:10:09 GMT"
},
{
"version": "v2",
"created": "Fri, 19 May 2017 00:24:43 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Günther",
"Manuel",
""
],
[
"Cruz",
"Steve",
""
],
[
"Rudd",
"Ethan M.",
""
],
[
"Boult",
"Terrance E.",
""
]
] | TITLE: Toward Open-Set Face Recognition
ABSTRACT: Much research has been conducted on both face identification and face
verification, with greater focus on the latter. Research on face identification
has mostly focused on using closed-set protocols, which assume that all probe
images used in evaluation contain identities of subjects that are enrolled in
the gallery. Real systems, however, where only a fraction of probe sample
identities are enrolled in the gallery, cannot make this closed-set assumption.
Instead, they must assume an open set of probe samples and be able to
reject/ignore those that correspond to unknown identities. In this paper, we
address the widespread misconception that thresholding verification-like scores
is a good way to solve the open-set face identification problem, by formulating
an open-set face identification protocol and evaluating different strategies
for assessing similarity. Our open-set identification protocol is based on the
canonical labeled faces in the wild (LFW) dataset. Additionally to the known
identities, we introduce the concepts of known unknowns (known, but
uninteresting persons) and unknown unknowns (people never seen before) to the
biometric community. We compare three algorithms for assessing similarity in a
deep feature space under an open-set protocol: thresholded verification-like
scores, linear discriminant analysis (LDA) scores, and an extreme value machine
(EVM) probabilities. Our findings suggest that thresholding EVM probabilities,
which are open-set by design, outperforms thresholding verification-like
scores.
| no_new_dataset | 0.955569 |
1705.03146 | Shiyang Yan | Shiyang Yan, Jeremy S. Smith, Wenjin Lu and Bailing Zhang | CHAM: action recognition using convolutional hierarchical attention
model | accepted by ICIP2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, the soft attention mechanism, which was originally proposed in
language processing, has been applied in computer vision tasks like image
captioning. This paper presents improvements to the soft attention model by
combining a convolutional LSTM with a hierarchical system architecture to
recognize action categories in videos. We call this model the Convolutional
Hierarchical Attention Model (CHAM). The model applies a convolutional
operation inside the LSTM cell and an attention map generation process to
recognize actions. The hierarchical architecture of this model is able to
explicitly reason on multi-granularities of action categories. The proposed
architecture achieved improved results on three publicly available datasets:
the UCF sports dataset, the Olympic sports dataset and the HMDB51 dataset.
| [
{
"version": "v1",
"created": "Tue, 9 May 2017 02:27:37 GMT"
},
{
"version": "v2",
"created": "Fri, 19 May 2017 06:11:26 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Yan",
"Shiyang",
""
],
[
"Smith",
"Jeremy S.",
""
],
[
"Lu",
"Wenjin",
""
],
[
"Zhang",
"Bailing",
""
]
] | TITLE: CHAM: action recognition using convolutional hierarchical attention
model
ABSTRACT: Recently, the soft attention mechanism, which was originally proposed in
language processing, has been applied in computer vision tasks like image
captioning. This paper presents improvements to the soft attention model by
combining a convolutional LSTM with a hierarchical system architecture to
recognize action categories in videos. We call this model the Convolutional
Hierarchical Attention Model (CHAM). The model applies a convolutional
operation inside the LSTM cell and an attention map generation process to
recognize actions. The hierarchical architecture of this model is able to
explicitly reason on multi-granularities of action categories. The proposed
architecture achieved improved results on three publicly available datasets:
the UCF sports dataset, the Olympic sports dataset and the HMDB51 dataset.
| no_new_dataset | 0.953966 |
1705.06753 | Konstantin Bauman | Evgeny Bauman, Konstantin Bauman | Discovering the Graph Structure in the Clustering Results | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In a standard cluster analysis, such as k-means, in addition to clusters
locations and distances between them, it's important to know if they are
connected or well separated from each other. The main focus of this paper is
discovering the relations between the resulting clusters. We propose a new
method which is based on pairwise overlapping k-means clustering, that in
addition to means of clusters provides the graph structure of their relations.
The proposed method has a set of parameters that can be tuned in order to
control the sensitivity of the model and the desired relative size of the
pairwise overlapping interval between means of two adjacent clusters, i.e.,
level of overlapping. We present the exact formula for calculating that
parameter. The empirical study presented in the paper demonstrates that our
approach works well not only on toy data but also compliments standard
clustering results with a reasonable graph structure on real datasets, such as
financial indices and restaurants.
| [
{
"version": "v1",
"created": "Thu, 18 May 2017 18:01:50 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Bauman",
"Evgeny",
""
],
[
"Bauman",
"Konstantin",
""
]
] | TITLE: Discovering the Graph Structure in the Clustering Results
ABSTRACT: In a standard cluster analysis, such as k-means, in addition to clusters
locations and distances between them, it's important to know if they are
connected or well separated from each other. The main focus of this paper is
discovering the relations between the resulting clusters. We propose a new
method which is based on pairwise overlapping k-means clustering, that in
addition to means of clusters provides the graph structure of their relations.
The proposed method has a set of parameters that can be tuned in order to
control the sensitivity of the model and the desired relative size of the
pairwise overlapping interval between means of two adjacent clusters, i.e.,
level of overlapping. We present the exact formula for calculating that
parameter. The empirical study presented in the paper demonstrates that our
approach works well not only on toy data but also compliments standard
clustering results with a reasonable graph structure on real datasets, such as
financial indices and restaurants.
| no_new_dataset | 0.949342 |
1705.06849 | Lianwen Jin | Songxuan Lai, Lianwen Jin, Weixin Yang | Online Signature Verification using Recurrent Neural Network and
Length-normalized Path Signature | 6 pages, 5 figures, 5 tables | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Inspired by the great success of recurrent neural networks (RNNs) in
sequential modeling, we introduce a novel RNN system to improve the performance
of online signature verification. The training objective is to directly
minimize intra-class variations and to push the distances between skilled
forgeries and genuine samples above a given threshold. By back-propagating the
training signals, our RNN network produced discriminative features with desired
metrics. Additionally, we propose a novel descriptor, called the
length-normalized path signature (LNPS), and apply it to online signature
verification. LNPS has interesting properties, such as scale invariance and
rotation invariance after linear combination, and shows promising results in
online signature verification. Experiments on the publicly available SVC-2004
dataset yielded state-of-the-art performance of 2.37% equal error rate (EER).
| [
{
"version": "v1",
"created": "Fri, 19 May 2017 02:27:58 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Lai",
"Songxuan",
""
],
[
"Jin",
"Lianwen",
""
],
[
"Yang",
"Weixin",
""
]
] | TITLE: Online Signature Verification using Recurrent Neural Network and
Length-normalized Path Signature
ABSTRACT: Inspired by the great success of recurrent neural networks (RNNs) in
sequential modeling, we introduce a novel RNN system to improve the performance
of online signature verification. The training objective is to directly
minimize intra-class variations and to push the distances between skilled
forgeries and genuine samples above a given threshold. By back-propagating the
training signals, our RNN network produced discriminative features with desired
metrics. Additionally, we propose a novel descriptor, called the
length-normalized path signature (LNPS), and apply it to online signature
verification. LNPS has interesting properties, such as scale invariance and
rotation invariance after linear combination, and shows promising results in
online signature verification. Experiments on the publicly available SVC-2004
dataset yielded state-of-the-art performance of 2.37% equal error rate (EER).
| no_new_dataset | 0.94699 |
1705.06871 | Shirui Li | You Hao, Shirui Li, Hanlin Mo, and Hua Li | Affine-Gradient Based Local Binary Pattern Descriptor for Texture
Classiffication | 11 pages,4 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a novel Affine-Gradient based Local Binary Pattern (AGLBP)
descriptor for texture classification. It is very hard to describe complicated
texture using single type information, such as Local Binary Pattern (LBP),
which just utilizes the sign information of the difference between the pixel
and its local neighbors. Our descriptor has three characteristics: 1) In order
to make full use of the information contained in the texture, the
Affine-Gradient, which is different from Euclidean-Gradient and invariant to
affine transformation is incorporated into AGLBP. 2) An improved method is
proposed for rotation invariance, which depends on the reference direction
calculating respect to local neighbors. 3) Feature selection method,
considering both the statistical frequency and the intraclass variance of the
training dataset, is also applied to reduce the dimensionality of descriptors.
Experiments on three standard texture datasets, Outex12, Outex10 and KTH-TIPS2,
are conducted to evaluate the performance of AGLBP. The results show that our
proposed descriptor gets better performance comparing to some state-of-the-art
rotation texture descriptors in texture classification.
| [
{
"version": "v1",
"created": "Fri, 19 May 2017 06:41:31 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Hao",
"You",
""
],
[
"Li",
"Shirui",
""
],
[
"Mo",
"Hanlin",
""
],
[
"Li",
"Hua",
""
]
] | TITLE: Affine-Gradient Based Local Binary Pattern Descriptor for Texture
Classiffication
ABSTRACT: We present a novel Affine-Gradient based Local Binary Pattern (AGLBP)
descriptor for texture classification. It is very hard to describe complicated
texture using single type information, such as Local Binary Pattern (LBP),
which just utilizes the sign information of the difference between the pixel
and its local neighbors. Our descriptor has three characteristics: 1) In order
to make full use of the information contained in the texture, the
Affine-Gradient, which is different from Euclidean-Gradient and invariant to
affine transformation is incorporated into AGLBP. 2) An improved method is
proposed for rotation invariance, which depends on the reference direction
calculating respect to local neighbors. 3) Feature selection method,
considering both the statistical frequency and the intraclass variance of the
training dataset, is also applied to reduce the dimensionality of descriptors.
Experiments on three standard texture datasets, Outex12, Outex10 and KTH-TIPS2,
are conducted to evaluate the performance of AGLBP. The results show that our
proposed descriptor gets better performance comparing to some state-of-the-art
rotation texture descriptors in texture classification.
| no_new_dataset | 0.949763 |
1705.06950 | Joao Carreira | Will Kay, Joao Carreira, Karen Simonyan, Brian Zhang, Chloe Hillier,
Sudheendra Vijayanarasimhan, Fabio Viola, Tim Green, Trevor Back, Paul
Natsev, Mustafa Suleyman and Andrew Zisserman | The Kinetics Human Action Video Dataset | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We describe the DeepMind Kinetics human action video dataset. The dataset
contains 400 human action classes, with at least 400 video clips for each
action. Each clip lasts around 10s and is taken from a different YouTube video.
The actions are human focussed and cover a broad range of classes including
human-object interactions such as playing instruments, as well as human-human
interactions such as shaking hands. We describe the statistics of the dataset,
how it was collected, and give some baseline performance figures for neural
network architectures trained and tested for human action classification on
this dataset. We also carry out a preliminary analysis of whether imbalance in
the dataset leads to bias in the classifiers.
| [
{
"version": "v1",
"created": "Fri, 19 May 2017 12:07:01 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Kay",
"Will",
""
],
[
"Carreira",
"Joao",
""
],
[
"Simonyan",
"Karen",
""
],
[
"Zhang",
"Brian",
""
],
[
"Hillier",
"Chloe",
""
],
[
"Vijayanarasimhan",
"Sudheendra",
""
],
[
"Viola",
"Fabio",
""
],
[
"Green",
"Tim",
""
],
[
"Back",
"Trevor",
""
],
[
"Natsev",
"Paul",
""
],
[
"Suleyman",
"Mustafa",
""
],
[
"Zisserman",
"Andrew",
""
]
] | TITLE: The Kinetics Human Action Video Dataset
ABSTRACT: We describe the DeepMind Kinetics human action video dataset. The dataset
contains 400 human action classes, with at least 400 video clips for each
action. Each clip lasts around 10s and is taken from a different YouTube video.
The actions are human focussed and cover a broad range of classes including
human-object interactions such as playing instruments, as well as human-human
interactions such as shaking hands. We describe the statistics of the dataset,
how it was collected, and give some baseline performance figures for neural
network architectures trained and tested for human action classification on
this dataset. We also carry out a preliminary analysis of whether imbalance in
the dataset leads to bias in the classifiers.
| new_dataset | 0.933975 |
1705.07008 | Leandro dos Santos | Leandro B. dos Santos, Magali S. Duran, Nathan S. Hartmann, Arnaldo
Candido Jr., Gustavo H. Paetzold, Sandra M. Aluisio | A Lightweight Regression Method to Infer Psycholinguistic Properties for
Brazilian Portuguese | Paper accepted for TSD2017 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Psycholinguistic properties of words have been used in various approaches to
Natural Language Processing tasks, such as text simplification and readability
assessment. Most of these properties are subjective, involving costly and
time-consuming surveys to be gathered. Recent approaches use the limited
datasets of psycholinguistic properties to extend them automatically to large
lexicons. However, some of the resources used by such approaches are not
available to most languages. This study presents a method to infer
psycholinguistic properties for Brazilian Portuguese (BP) using regressors
built with a light set of features usually available for less resourced
languages: word length, frequency lists, lexical databases composed of school
dictionaries and word embedding models. The correlations between the properties
inferred are close to those obtained by related works. The resulting resource
contains 26,874 words in BP annotated with concreteness, age of acquisition,
imageability and subjective frequency.
| [
{
"version": "v1",
"created": "Fri, 19 May 2017 14:17:31 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Santos",
"Leandro B. dos",
""
],
[
"Duran",
"Magali S.",
""
],
[
"Hartmann",
"Nathan S.",
""
],
[
"Candido",
"Arnaldo",
"Jr."
],
[
"Paetzold",
"Gustavo H.",
""
],
[
"Aluisio",
"Sandra M.",
""
]
] | TITLE: A Lightweight Regression Method to Infer Psycholinguistic Properties for
Brazilian Portuguese
ABSTRACT: Psycholinguistic properties of words have been used in various approaches to
Natural Language Processing tasks, such as text simplification and readability
assessment. Most of these properties are subjective, involving costly and
time-consuming surveys to be gathered. Recent approaches use the limited
datasets of psycholinguistic properties to extend them automatically to large
lexicons. However, some of the resources used by such approaches are not
available to most languages. This study presents a method to infer
psycholinguistic properties for Brazilian Portuguese (BP) using regressors
built with a light set of features usually available for less resourced
languages: word length, frequency lists, lexical databases composed of school
dictionaries and word embedding models. The correlations between the properties
inferred are close to those obtained by related works. The resulting resource
contains 26,874 words in BP annotated with concreteness, age of acquisition,
imageability and subjective frequency.
| no_new_dataset | 0.944434 |
1705.07015 | Jen-Wei Kuo | Jen-wei Kuo, Jonathan Mamou, Yao Wang, Emi Saegusa-Beecroft, Junji
Machi, and Ernest J. Feleppa | Segmentation of 3D High-frequency Ultrasound Images of Human Lymph Nodes
Using Graph Cut with Energy Functional Adapted to Local Intensity
Distribution | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Previous studies by our group have shown that three-dimensional
high-frequency quantitative ultrasound methods have the potential to
differentiate metastatic lymph nodes from cancer-free lymph nodes dissected
from human cancer patients. To successfully perform these methods inside the
lymph node parenchyma, an automatic segmentation method is highly desired to
exclude the surrounding thin layer of fat from quantitative ultrasound
processing and accurately correct for ultrasound attenuation. In high-frequency
ultrasound images of lymph nodes, the intensity distribution of lymph node
parenchyma and fat varies spatially because of acoustic attenuation and
focusing effects. Thus, the intensity contrast between two object regions
(e.g., lymph node parenchyma and fat) is also spatially varying. In our
previous work, nested graph cut demonstrated its ability to simultaneously
segment lymph node parenchyma, fat, and the outer phosphate-buffered saline
bath even when some boundaries are lost because of acoustic attenuation and
focusing effects. This paper describes a novel approach called graph cut with
locally adaptive energy to further deal with spatially varying distributions of
lymph node parenchyma and fat caused by inhomogeneous acoustic attenuation. The
proposed method achieved Dice similarity coefficients of 0.937+-0.035 when
compared to expert manual segmentation on a representative dataset consisting
of 115 three-dimensional lymph node images obtained from colorectal cancer
patients.
| [
{
"version": "v1",
"created": "Fri, 19 May 2017 14:25:20 GMT"
}
] | 2017-05-22T00:00:00 | [
[
"Kuo",
"Jen-wei",
""
],
[
"Mamou",
"Jonathan",
""
],
[
"Wang",
"Yao",
""
],
[
"Saegusa-Beecroft",
"Emi",
""
],
[
"Machi",
"Junji",
""
],
[
"Feleppa",
"Ernest J.",
""
]
] | TITLE: Segmentation of 3D High-frequency Ultrasound Images of Human Lymph Nodes
Using Graph Cut with Energy Functional Adapted to Local Intensity
Distribution
ABSTRACT: Previous studies by our group have shown that three-dimensional
high-frequency quantitative ultrasound methods have the potential to
differentiate metastatic lymph nodes from cancer-free lymph nodes dissected
from human cancer patients. To successfully perform these methods inside the
lymph node parenchyma, an automatic segmentation method is highly desired to
exclude the surrounding thin layer of fat from quantitative ultrasound
processing and accurately correct for ultrasound attenuation. In high-frequency
ultrasound images of lymph nodes, the intensity distribution of lymph node
parenchyma and fat varies spatially because of acoustic attenuation and
focusing effects. Thus, the intensity contrast between two object regions
(e.g., lymph node parenchyma and fat) is also spatially varying. In our
previous work, nested graph cut demonstrated its ability to simultaneously
segment lymph node parenchyma, fat, and the outer phosphate-buffered saline
bath even when some boundaries are lost because of acoustic attenuation and
focusing effects. This paper describes a novel approach called graph cut with
locally adaptive energy to further deal with spatially varying distributions of
lymph node parenchyma and fat caused by inhomogeneous acoustic attenuation. The
proposed method achieved Dice similarity coefficients of 0.937+-0.035 when
compared to expert manual segmentation on a representative dataset consisting
of 115 three-dimensional lymph node images obtained from colorectal cancer
patients.
| new_dataset | 0.966156 |
1611.05125 | Paritosh Parmar | Paritosh Parmar and Brendan Tran Morris | Learning To Score Olympic Events | CVPR 2017 - CVSports Workshop | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Estimating action quality, the process of assigning a "score" to the
execution of an action, is crucial in areas such as sports and health care.
Unlike action recognition, which has millions of examples to learn from, the
action quality datasets that are currently available are small -- typically
comprised of only a few hundred samples. This work presents three frameworks
for evaluating Olympic sports which utilize spatiotemporal features learned
using 3D convolutional neural networks (C3D) and perform score regression with
i) SVR, ii) LSTM, and iii) LSTM followed by SVR. An efficient training
mechanism for the limited data scenarios is presented for clip-based training
with LSTM. The proposed systems show significant improvement over existing
quality assessment approaches on the task of predicting scores of Olympic
events {diving, vault, figure skating}. While the SVR-based frameworks yield
better results, LSTM-based frameworks are more natural for describing an action
and can be used for improvement feedback.
| [
{
"version": "v1",
"created": "Wed, 16 Nov 2016 02:56:24 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jan 2017 00:47:29 GMT"
},
{
"version": "v3",
"created": "Thu, 18 May 2017 05:55:24 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Parmar",
"Paritosh",
""
],
[
"Morris",
"Brendan Tran",
""
]
] | TITLE: Learning To Score Olympic Events
ABSTRACT: Estimating action quality, the process of assigning a "score" to the
execution of an action, is crucial in areas such as sports and health care.
Unlike action recognition, which has millions of examples to learn from, the
action quality datasets that are currently available are small -- typically
comprised of only a few hundred samples. This work presents three frameworks
for evaluating Olympic sports which utilize spatiotemporal features learned
using 3D convolutional neural networks (C3D) and perform score regression with
i) SVR, ii) LSTM, and iii) LSTM followed by SVR. An efficient training
mechanism for the limited data scenarios is presented for clip-based training
with LSTM. The proposed systems show significant improvement over existing
quality assessment approaches on the task of predicting scores of Olympic
events {diving, vault, figure skating}. While the SVR-based frameworks yield
better results, LSTM-based frameworks are more natural for describing an action
and can be used for improvement feedback.
| no_new_dataset | 0.947527 |
1704.02788 | Chuanqi Tan | Chuanqi Tan, Furu Wei, Pengjie Ren, Weifeng Lv, Ming Zhou | Entity Linking for Queries by Searching Wikipedia Sentences | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a simple yet effective approach for linking entities in queries.
The key idea is to search sentences similar to a query from Wikipedia articles
and directly use the human-annotated entities in the similar sentences as
candidate entities for the query. Then, we employ a rich set of features, such
as link-probability, context-matching, word embeddings, and relatedness among
candidate entities as well as their related entities, to rank the candidates
under a regression based framework. The advantages of our approach lie in two
aspects, which contribute to the ranking process and final linking result.
First, it can greatly reduce the number of candidate entities by filtering out
irrelevant entities with the words in the query. Second, we can obtain the
query sensitive prior probability in addition to the static link-probability
derived from all Wikipedia articles. We conduct experiments on two benchmark
datasets on entity linking for queries, namely the ERD14 dataset and the GERDAQ
dataset. Experimental results show that our method outperforms state-of-the-art
systems and yields 75.0% in F1 on the ERD14 dataset and 56.9% on the GERDAQ
dataset.
| [
{
"version": "v1",
"created": "Mon, 10 Apr 2017 10:19:53 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Apr 2017 06:59:56 GMT"
},
{
"version": "v3",
"created": "Thu, 18 May 2017 08:03:49 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Tan",
"Chuanqi",
""
],
[
"Wei",
"Furu",
""
],
[
"Ren",
"Pengjie",
""
],
[
"Lv",
"Weifeng",
""
],
[
"Zhou",
"Ming",
""
]
] | TITLE: Entity Linking for Queries by Searching Wikipedia Sentences
ABSTRACT: We present a simple yet effective approach for linking entities in queries.
The key idea is to search sentences similar to a query from Wikipedia articles
and directly use the human-annotated entities in the similar sentences as
candidate entities for the query. Then, we employ a rich set of features, such
as link-probability, context-matching, word embeddings, and relatedness among
candidate entities as well as their related entities, to rank the candidates
under a regression based framework. The advantages of our approach lie in two
aspects, which contribute to the ranking process and final linking result.
First, it can greatly reduce the number of candidate entities by filtering out
irrelevant entities with the words in the query. Second, we can obtain the
query sensitive prior probability in addition to the static link-probability
derived from all Wikipedia articles. We conduct experiments on two benchmark
datasets on entity linking for queries, namely the ERD14 dataset and the GERDAQ
dataset. Experimental results show that our method outperforms state-of-the-art
systems and yields 75.0% in F1 on the ERD14 dataset and 56.9% on the GERDAQ
dataset.
| no_new_dataset | 0.951369 |
1705.03261 | Zibo Yi | Zibo Yi, Shasha Li, Jie Yu, Qingbo Wu | Drug-drug Interaction Extraction via Recurrent Neural Network with
Multiple Attention Layers | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Drug-drug interaction (DDI) is a vital information when physicians and
pharmacists intend to co-administer two or more drugs. Thus, several DDI
databases are constructed to avoid mistakenly combined use. In recent years,
automatically extracting DDIs from biomedical text has drawn researchers'
attention. However, the existing work utilize either complex feature
engineering or NLP tools, both of which are insufficient for sentence
comprehension. Inspired by the deep learning approaches in natural language
processing, we propose a recur- rent neural network model with multiple
attention layers for DDI classification. We evaluate our model on 2013 SemEval
DDIExtraction dataset. The experiments show that our model classifies most of
the drug pairs into correct DDI categories, which outperforms the existing NLP
or deep learning methods.
| [
{
"version": "v1",
"created": "Tue, 9 May 2017 10:22:48 GMT"
},
{
"version": "v2",
"created": "Thu, 18 May 2017 15:54:36 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Yi",
"Zibo",
""
],
[
"Li",
"Shasha",
""
],
[
"Yu",
"Jie",
""
],
[
"Wu",
"Qingbo",
""
]
] | TITLE: Drug-drug Interaction Extraction via Recurrent Neural Network with
Multiple Attention Layers
ABSTRACT: Drug-drug interaction (DDI) is a vital information when physicians and
pharmacists intend to co-administer two or more drugs. Thus, several DDI
databases are constructed to avoid mistakenly combined use. In recent years,
automatically extracting DDIs from biomedical text has drawn researchers'
attention. However, the existing work utilize either complex feature
engineering or NLP tools, both of which are insufficient for sentence
comprehension. Inspired by the deep learning approaches in natural language
processing, we propose a recur- rent neural network model with multiple
attention layers for DDI classification. We evaluate our model on 2013 SemEval
DDIExtraction dataset. The experiments show that our model classifies most of
the drug pairs into correct DDI categories, which outperforms the existing NLP
or deep learning methods.
| no_new_dataset | 0.945197 |
1705.06362 | Darvin Yi | Darvin Yi, Rebecca Lynn Sawyer, David Cohn III, Jared Dunnmon, Carson
Lam, Xuerong Xiao, and Daniel Rubin | Optimizing and Visualizing Deep Learning for Benign/Malignant
Classification in Breast Tumors | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Breast cancer has the highest incidence and second highest mortality rate for
women in the US. Our study aims to utilize deep learning for benign/malignant
classification of mammogram tumors using a subset of cases from the Digital
Database of Screening Mammography (DDSM). Though it was a small dataset from
the view of Deep Learning (about 1000 patients), we show that currently state
of the art architectures of deep learning can find a robust signal, even when
trained from scratch. Using convolutional neural networks (CNNs), we are able
to achieve an accuracy of 85% and an ROC AUC of 0.91, while leading
hand-crafted feature based methods are only able to achieve an accuracy of 71%.
We investigate an amalgamation of architectures to show that our best result is
reached with an ensemble of the lightweight GoogLe Nets tasked with
interpreting both the coronal caudal view and the mediolateral oblique view,
simply averaging the probability scores of both views to make the final
prediction. In addition, we have created a novel method to visualize what
features the neural network detects for the benign/malignant classification,
and have correlated those features with well known radiological features, such
as spiculation. Our algorithm significantly improves existing classification
methods for mammography lesions and identifies features that correlate with
established clinical markers.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 22:35:28 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Yi",
"Darvin",
""
],
[
"Sawyer",
"Rebecca Lynn",
""
],
[
"Cohn",
"David",
"III"
],
[
"Dunnmon",
"Jared",
""
],
[
"Lam",
"Carson",
""
],
[
"Xiao",
"Xuerong",
""
],
[
"Rubin",
"Daniel",
""
]
] | TITLE: Optimizing and Visualizing Deep Learning for Benign/Malignant
Classification in Breast Tumors
ABSTRACT: Breast cancer has the highest incidence and second highest mortality rate for
women in the US. Our study aims to utilize deep learning for benign/malignant
classification of mammogram tumors using a subset of cases from the Digital
Database of Screening Mammography (DDSM). Though it was a small dataset from
the view of Deep Learning (about 1000 patients), we show that currently state
of the art architectures of deep learning can find a robust signal, even when
trained from scratch. Using convolutional neural networks (CNNs), we are able
to achieve an accuracy of 85% and an ROC AUC of 0.91, while leading
hand-crafted feature based methods are only able to achieve an accuracy of 71%.
We investigate an amalgamation of architectures to show that our best result is
reached with an ensemble of the lightweight GoogLe Nets tasked with
interpreting both the coronal caudal view and the mediolateral oblique view,
simply averaging the probability scores of both views to make the final
prediction. In addition, we have created a novel method to visualize what
features the neural network detects for the benign/malignant classification,
and have correlated those features with well known radiological features, such
as spiculation. Our algorithm significantly improves existing classification
methods for mammography lesions and identifies features that correlate with
established clinical markers.
| no_new_dataset | 0.945045 |
1705.06371 | Robert Durrant | Xianghui Luo and Robert J. Durrant | Maximum Margin Principal Components | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Principal Component Analysis (PCA) is a very successful dimensionality
reduction technique, widely used in predictive modeling. A key factor in its
widespread use in this domain is the fact that the projection of a dataset onto
its first $K$ principal components minimizes the sum of squared errors between
the original data and the projected data over all possible rank $K$
projections. Thus, PCA provides optimal low-rank representations of data for
least-squares linear regression under standard modeling assumptions. On the
other hand, when the loss function for a prediction problem is not the
least-squares error, PCA is typically a heuristic choice of dimensionality
reduction -- in particular for classification problems under the zero-one loss.
In this paper we target classification problems by proposing a straightforward
alternative to PCA that aims to minimize the difference in margin distribution
between the original and the projected data. Extensive experiments show that
our simple approach typically outperforms PCA on any particular dataset, in
terms of classification error, though this difference is not always
statistically significant, and despite being a filter method is frequently
competitive with Partial Least Squares (PLS) and Lasso on a wide range of
datasets.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 23:45:11 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Luo",
"Xianghui",
""
],
[
"Durrant",
"Robert J.",
""
]
] | TITLE: Maximum Margin Principal Components
ABSTRACT: Principal Component Analysis (PCA) is a very successful dimensionality
reduction technique, widely used in predictive modeling. A key factor in its
widespread use in this domain is the fact that the projection of a dataset onto
its first $K$ principal components minimizes the sum of squared errors between
the original data and the projected data over all possible rank $K$
projections. Thus, PCA provides optimal low-rank representations of data for
least-squares linear regression under standard modeling assumptions. On the
other hand, when the loss function for a prediction problem is not the
least-squares error, PCA is typically a heuristic choice of dimensionality
reduction -- in particular for classification problems under the zero-one loss.
In this paper we target classification problems by proposing a straightforward
alternative to PCA that aims to minimize the difference in margin distribution
between the original and the projected data. Extensive experiments show that
our simple approach typically outperforms PCA on any particular dataset, in
terms of classification error, though this difference is not always
statistically significant, and despite being a filter method is frequently
competitive with Partial Least Squares (PLS) and Lasso on a wide range of
datasets.
| no_new_dataset | 0.942876 |
1705.06516 | Pedro F. Proen\c{c}a | Pedro F. Proen\c{c}a and Yang Gao | Probabilistic Combination of Noisy Points and Planes for RGB-D Odometry | Accepted to TAROS 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work proposes a visual odometry method that combines points and plane
primitives, extracted from a noisy depth camera. Depth measurement uncertainty
is modelled and propagated through the extraction of geometric primitives to
the frame-to-frame motion estimation, where pose is optimized by weighting the
residuals of 3D point and planes matches, according to their uncertainties.
Results on an RGB-D dataset show that the combination of points and planes,
through the proposed method, is able to perform well in poorly textured
environments, where point-based odometry is bound to fail.
| [
{
"version": "v1",
"created": "Thu, 18 May 2017 10:53:51 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Proença",
"Pedro F.",
""
],
[
"Gao",
"Yang",
""
]
] | TITLE: Probabilistic Combination of Noisy Points and Planes for RGB-D Odometry
ABSTRACT: This work proposes a visual odometry method that combines points and plane
primitives, extracted from a noisy depth camera. Depth measurement uncertainty
is modelled and propagated through the extraction of geometric primitives to
the frame-to-frame motion estimation, where pose is optimized by weighting the
residuals of 3D point and planes matches, according to their uncertainties.
Results on an RGB-D dataset show that the combination of points and planes,
through the proposed method, is able to perform well in poorly textured
environments, where point-based odometry is bound to fail.
| no_new_dataset | 0.944842 |
1705.06560 | Kuo-Hao Zeng | Kuo-Hao Zeng, Shih-Han Chou, Fu-Hsiang Chan, Juan Carlos Niebles, Min
Sun | Agent-Centric Risk Assessment: Accident Anticipation and Risky Region
Localization | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For survival, a living agent must have the ability to assess risk (1) by
temporally anticipating accidents before they occur, and (2) by spatially
localizing risky regions in the environment to move away from threats. In this
paper, we take an agent-centric approach to study the accident anticipation and
risky region localization tasks. We propose a novel soft-attention Recurrent
Neural Network (RNN) which explicitly models both spatial and appearance-wise
non-linear interaction between the agent triggering the event and another agent
or static-region involved. In order to test our proposed method, we introduce
the Epic Fail (EF) dataset consisting of 3000 viral videos capturing various
accidents. In the experiments, we evaluate the risk assessment accuracy both in
the temporal domain (accident anticipation) and spatial domain (risky region
localization) on our EF dataset and the Street Accident (SA) dataset. Our
method consistently outperforms other baselines on both datasets.
| [
{
"version": "v1",
"created": "Thu, 18 May 2017 12:56:20 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Zeng",
"Kuo-Hao",
""
],
[
"Chou",
"Shih-Han",
""
],
[
"Chan",
"Fu-Hsiang",
""
],
[
"Niebles",
"Juan Carlos",
""
],
[
"Sun",
"Min",
""
]
] | TITLE: Agent-Centric Risk Assessment: Accident Anticipation and Risky Region
Localization
ABSTRACT: For survival, a living agent must have the ability to assess risk (1) by
temporally anticipating accidents before they occur, and (2) by spatially
localizing risky regions in the environment to move away from threats. In this
paper, we take an agent-centric approach to study the accident anticipation and
risky region localization tasks. We propose a novel soft-attention Recurrent
Neural Network (RNN) which explicitly models both spatial and appearance-wise
non-linear interaction between the agent triggering the event and another agent
or static-region involved. In order to test our proposed method, we introduce
the Epic Fail (EF) dataset consisting of 3000 viral videos capturing various
accidents. In the experiments, we evaluate the risk assessment accuracy both in
the temporal domain (accident anticipation) and spatial domain (risky region
localization) on our EF dataset and the Street Accident (SA) dataset. Our
method consistently outperforms other baselines on both datasets.
| new_dataset | 0.958226 |
1705.06599 | Boyue Wang | Boyue Wang, Yongli Hu, Junbin Gao, Yanfeng Sun and Baocai Yin | Localized LRR on Grassmann Manifolds: An Extrinsic View | IEEE Transactions on Circuits and Systems for Video Technology with
Minor Revisions. arXiv admin note: text overlap with arXiv:1504.01807 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Subspace data representation has recently become a common practice in many
computer vision tasks. It demands generalizing classical machine learning
algorithms for subspace data. Low-Rank Representation (LRR) is one of the most
successful models for clustering vectorial data according to their subspace
structures. This paper explores the possibility of extending LRR for subspace
data on Grassmann manifolds. Rather than directly embedding the Grassmann
manifolds into the symmetric matrix space, an extrinsic view is taken to build
the LRR self-representation in the local area of the tangent space at each
Grassmannian point, resulting in a localized LRR method on Grassmann manifolds.
A novel algorithm for solving the proposed model is investigated and
implemented. The performance of the new clustering algorithm is assessed
through experiments on several real-world datasets including MNIST handwritten
digits, ballet video clips, SKIG action clips, DynTex++ dataset and highway
traffic video clips. The experimental results show the new method outperforms a
number of state-of-the-art clustering methods
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 03:04:43 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Wang",
"Boyue",
""
],
[
"Hu",
"Yongli",
""
],
[
"Gao",
"Junbin",
""
],
[
"Sun",
"Yanfeng",
""
],
[
"Yin",
"Baocai",
""
]
] | TITLE: Localized LRR on Grassmann Manifolds: An Extrinsic View
ABSTRACT: Subspace data representation has recently become a common practice in many
computer vision tasks. It demands generalizing classical machine learning
algorithms for subspace data. Low-Rank Representation (LRR) is one of the most
successful models for clustering vectorial data according to their subspace
structures. This paper explores the possibility of extending LRR for subspace
data on Grassmann manifolds. Rather than directly embedding the Grassmann
manifolds into the symmetric matrix space, an extrinsic view is taken to build
the LRR self-representation in the local area of the tangent space at each
Grassmannian point, resulting in a localized LRR method on Grassmann manifolds.
A novel algorithm for solving the proposed model is investigated and
implemented. The performance of the new clustering algorithm is assessed
through experiments on several real-world datasets including MNIST handwritten
digits, ballet video clips, SKIG action clips, DynTex++ dataset and highway
traffic video clips. The experimental results show the new method outperforms a
number of state-of-the-art clustering methods
| no_new_dataset | 0.950457 |
1705.06687 | Nick Johnston | Michele Covell, Nick Johnston, David Minnen, Sung Jin Hwang, Joel
Shor, Saurabh Singh, Damien Vincent, George Toderici | Target-Quality Image Compression with Recurrent, Convolutional Neural
Networks | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a stop-code tolerant (SCT) approach to training recurrent
convolutional neural networks for lossy image compression. Our methods
introduce a multi-pass training method to combine the training goals of
high-quality reconstructions in areas around stop-code masking as well as in
highly-detailed areas. These methods lead to lower true bitrates for a given
recursion count, both pre- and post-entropy coding, even using unstructured
LZ77 code compression. The pre-LZ77 gains are achieved by trimming stop codes.
The post-LZ77 gains are due to the highly unequal distributions of 0/1 codes
from the SCT architectures. With these code compressions, the SCT architecture
maintains or exceeds the image quality at all compression rates compared to
JPEG and to RNN auto-encoders across the Kodak dataset. In addition, the SCT
coding results in lower variance in image quality across the extent of the
image, a characteristic that has been shown to be important in human ratings of
image quality
| [
{
"version": "v1",
"created": "Thu, 18 May 2017 16:44:31 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Covell",
"Michele",
""
],
[
"Johnston",
"Nick",
""
],
[
"Minnen",
"David",
""
],
[
"Hwang",
"Sung Jin",
""
],
[
"Shor",
"Joel",
""
],
[
"Singh",
"Saurabh",
""
],
[
"Vincent",
"Damien",
""
],
[
"Toderici",
"George",
""
]
] | TITLE: Target-Quality Image Compression with Recurrent, Convolutional Neural
Networks
ABSTRACT: We introduce a stop-code tolerant (SCT) approach to training recurrent
convolutional neural networks for lossy image compression. Our methods
introduce a multi-pass training method to combine the training goals of
high-quality reconstructions in areas around stop-code masking as well as in
highly-detailed areas. These methods lead to lower true bitrates for a given
recursion count, both pre- and post-entropy coding, even using unstructured
LZ77 code compression. The pre-LZ77 gains are achieved by trimming stop codes.
The post-LZ77 gains are due to the highly unequal distributions of 0/1 codes
from the SCT architectures. With these code compressions, the SCT architecture
maintains or exceeds the image quality at all compression rates compared to
JPEG and to RNN auto-encoders across the Kodak dataset. In addition, the SCT
coding results in lower variance in image quality across the extent of the
image, a characteristic that has been shown to be important in human ratings of
image quality
| no_new_dataset | 0.942823 |
1705.06709 | Zhuolin Jiang | Zhuolin Jiang, Viktor Rozgic, Sancar Adali | Learning Spatiotemporal Features for Infrared Action Recognition with 3D
Convolutional Neural Networks | null | null | null | null | cs.CV cs.AI cs.LG cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Infrared (IR) imaging has the potential to enable more robust action
recognition systems compared to visible spectrum cameras due to lower
sensitivity to lighting conditions and appearance variability. While the action
recognition task on videos collected from visible spectrum imaging has received
much attention, action recognition in IR videos is significantly less explored.
Our objective is to exploit imaging data in this modality for the action
recognition task. In this work, we propose a novel two-stream 3D convolutional
neural network (CNN) architecture by introducing the discriminative code layer
and the corresponding discriminative code loss function. The proposed network
processes IR image and the IR-based optical flow field sequences. We pretrain
the 3D CNN model on the visible spectrum Sports-1M action dataset and finetune
it on the Infrared Action Recognition (InfAR) dataset. To our best knowledge,
this is the first application of the 3D CNN to action recognition in the IR
domain. We conduct an elaborate analysis of different fusion schemes (weighted
average, single and double-layer neural nets) applied to different 3D CNN
outputs. Experimental results demonstrate that our approach can achieve
state-of-the-art average precision (AP) performances on the InfAR dataset: (1)
the proposed two-stream 3D CNN achieves the best reported 77.5% AP, and (2) our
3D CNN model applied to the optical flow fields achieves the best reported
single stream 75.42% AP.
| [
{
"version": "v1",
"created": "Thu, 18 May 2017 17:26:34 GMT"
}
] | 2017-05-19T00:00:00 | [
[
"Jiang",
"Zhuolin",
""
],
[
"Rozgic",
"Viktor",
""
],
[
"Adali",
"Sancar",
""
]
] | TITLE: Learning Spatiotemporal Features for Infrared Action Recognition with 3D
Convolutional Neural Networks
ABSTRACT: Infrared (IR) imaging has the potential to enable more robust action
recognition systems compared to visible spectrum cameras due to lower
sensitivity to lighting conditions and appearance variability. While the action
recognition task on videos collected from visible spectrum imaging has received
much attention, action recognition in IR videos is significantly less explored.
Our objective is to exploit imaging data in this modality for the action
recognition task. In this work, we propose a novel two-stream 3D convolutional
neural network (CNN) architecture by introducing the discriminative code layer
and the corresponding discriminative code loss function. The proposed network
processes IR image and the IR-based optical flow field sequences. We pretrain
the 3D CNN model on the visible spectrum Sports-1M action dataset and finetune
it on the Infrared Action Recognition (InfAR) dataset. To our best knowledge,
this is the first application of the 3D CNN to action recognition in the IR
domain. We conduct an elaborate analysis of different fusion schemes (weighted
average, single and double-layer neural nets) applied to different 3D CNN
outputs. Experimental results demonstrate that our approach can achieve
state-of-the-art average precision (AP) performances on the InfAR dataset: (1)
the proposed two-stream 3D CNN achieves the best reported 77.5% AP, and (2) our
3D CNN model applied to the optical flow fields achieves the best reported
single stream 75.42% AP.
| no_new_dataset | 0.952175 |
1509.07831 | Jaeyong Sung | Jaeyong Sung, Ian Lenz, Ashutosh Saxena | Deep Multimodal Embedding: Manipulating Novel Objects with Point-clouds,
Language and Trajectories | IEEE International Conference on Robotics and Automation (ICRA), 2017 | null | null | null | cs.RO cs.AI cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A robot operating in a real-world environment needs to perform reasoning over
a variety of sensor modalities such as vision, language and motion
trajectories. However, it is extremely challenging to manually design features
relating such disparate modalities. In this work, we introduce an algorithm
that learns to embed point-cloud, natural language, and manipulation trajectory
data into a shared embedding space with a deep neural network. To learn
semantically meaningful spaces throughout our network, we use a loss-based
margin to bring embeddings of relevant pairs closer together while driving
less-relevant cases from different modalities further apart. We use this both
to pre-train its lower layers and fine-tune our final embedding space, leading
to a more robust representation. We test our algorithm on the task of
manipulating novel objects and appliances based on prior experience with other
objects. On a large dataset, we achieve significant improvements in both
accuracy and inference time over the previous state of the art. We also perform
end-to-end experiments on a PR2 robot utilizing our learned embedding space.
| [
{
"version": "v1",
"created": "Fri, 25 Sep 2015 18:55:45 GMT"
},
{
"version": "v2",
"created": "Wed, 17 May 2017 15:12:33 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Sung",
"Jaeyong",
""
],
[
"Lenz",
"Ian",
""
],
[
"Saxena",
"Ashutosh",
""
]
] | TITLE: Deep Multimodal Embedding: Manipulating Novel Objects with Point-clouds,
Language and Trajectories
ABSTRACT: A robot operating in a real-world environment needs to perform reasoning over
a variety of sensor modalities such as vision, language and motion
trajectories. However, it is extremely challenging to manually design features
relating such disparate modalities. In this work, we introduce an algorithm
that learns to embed point-cloud, natural language, and manipulation trajectory
data into a shared embedding space with a deep neural network. To learn
semantically meaningful spaces throughout our network, we use a loss-based
margin to bring embeddings of relevant pairs closer together while driving
less-relevant cases from different modalities further apart. We use this both
to pre-train its lower layers and fine-tune our final embedding space, leading
to a more robust representation. We test our algorithm on the task of
manipulating novel objects and appliances based on prior experience with other
objects. On a large dataset, we achieve significant improvements in both
accuracy and inference time over the previous state of the art. We also perform
end-to-end experiments on a PR2 robot utilizing our learned embedding space.
| no_new_dataset | 0.944382 |
1603.08308 | Minsu Park | Minsu Park, Mor Naaman, Jonah Berger | A Data-driven Study of View Duration on YouTube | 4 pages, 2 tables, Accepted to the 10th International AAAI Conference
on Web and Social Media, ICWSM'16 | 10th International AAAI Conference on Web and Social Media (ICWSM
2016) | null | null | cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Video watching had emerged as one of the most frequent media activities on
the Internet. Yet, little is known about how users watch online video. Using
two distinct YouTube datasets, a set of random YouTube videos crawled from the
Web and a set of videos watched by participants tracked by a Chrome extension,
we examine whether and how indicators of collective preferences and reactions
are associated with view duration of videos. We show that video view duration
is positively associated with the video's view count, the number of likes per
view, and the negative sentiment in the comments. These metrics and reactions
have a significant predictive power over the duration the video is watched by
individuals. Our findings provide a more precise understandings of user
engagement with video content in social media beyond view count.
| [
{
"version": "v1",
"created": "Mon, 28 Mar 2016 04:55:21 GMT"
},
{
"version": "v2",
"created": "Tue, 29 Mar 2016 15:46:07 GMT"
},
{
"version": "v3",
"created": "Wed, 17 May 2017 04:27:20 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Park",
"Minsu",
""
],
[
"Naaman",
"Mor",
""
],
[
"Berger",
"Jonah",
""
]
] | TITLE: A Data-driven Study of View Duration on YouTube
ABSTRACT: Video watching had emerged as one of the most frequent media activities on
the Internet. Yet, little is known about how users watch online video. Using
two distinct YouTube datasets, a set of random YouTube videos crawled from the
Web and a set of videos watched by participants tracked by a Chrome extension,
we examine whether and how indicators of collective preferences and reactions
are associated with view duration of videos. We show that video view duration
is positively associated with the video's view count, the number of likes per
view, and the negative sentiment in the comments. These metrics and reactions
have a significant predictive power over the duration the video is watched by
individuals. Our findings provide a more precise understandings of user
engagement with video content in social media beyond view count.
| no_new_dataset | 0.912903 |
1604.02071 | Reinhard Heckel | Reinhard Heckel, Michail Vlachos, Thomas Parnell, and Celestine
D\"unner | Scalable and interpretable product recommendations via overlapping
co-clustering | In IEEE International Conference on Data Engineering (ICDE) 2017 | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the problem of generating interpretable recommendations by
identifying overlapping co-clusters of clients and products, based only on
positive or implicit feedback. Our approach is applicable on very large
datasets because it exhibits almost linear complexity in the input examples and
the number of co-clusters. We show, both on real industrial data and on
publicly available datasets, that the recommendation accuracy of our algorithm
is competitive to that of state-of-art matrix factorization techniques. In
addition, our technique has the advantage of offering recommendations that are
textually and visually interpretable. Finally, we examine how to implement our
technique efficiently on Graphical Processing Units (GPUs).
| [
{
"version": "v1",
"created": "Thu, 7 Apr 2016 16:40:53 GMT"
},
{
"version": "v2",
"created": "Wed, 17 May 2017 17:58:51 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Heckel",
"Reinhard",
""
],
[
"Vlachos",
"Michail",
""
],
[
"Parnell",
"Thomas",
""
],
[
"Dünner",
"Celestine",
""
]
] | TITLE: Scalable and interpretable product recommendations via overlapping
co-clustering
ABSTRACT: We consider the problem of generating interpretable recommendations by
identifying overlapping co-clusters of clients and products, based only on
positive or implicit feedback. Our approach is applicable on very large
datasets because it exhibits almost linear complexity in the input examples and
the number of co-clusters. We show, both on real industrial data and on
publicly available datasets, that the recommendation accuracy of our algorithm
is competitive to that of state-of-art matrix factorization techniques. In
addition, our technique has the advantage of offering recommendations that are
textually and visually interpretable. Finally, we examine how to implement our
technique efficiently on Graphical Processing Units (GPUs).
| no_new_dataset | 0.945147 |
1611.09957 | Ehsan Amid | Ehsan Amid, Nikos Vlassis, Manfred K. Warmuth | Low-dimensional Data Embedding via Robust Ranking | null | null | null | null | cs.AI cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We describe a new method called t-ETE for finding a low-dimensional embedding
of a set of objects in Euclidean space. We formulate the embedding problem as a
joint ranking problem over a set of triplets, where each triplet captures the
relative similarities between three objects in the set. By exploiting recent
advances in robust ranking, t-ETE produces high-quality embeddings even in the
presence of a significant amount of noise and better preserves local scale than
known methods, such as t-STE and t-SNE. In particular, our method produces
significantly better results than t-SNE on signature datasets while also being
faster to compute.
| [
{
"version": "v1",
"created": "Wed, 30 Nov 2016 01:03:11 GMT"
},
{
"version": "v2",
"created": "Tue, 16 May 2017 21:21:03 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Amid",
"Ehsan",
""
],
[
"Vlassis",
"Nikos",
""
],
[
"Warmuth",
"Manfred K.",
""
]
] | TITLE: Low-dimensional Data Embedding via Robust Ranking
ABSTRACT: We describe a new method called t-ETE for finding a low-dimensional embedding
of a set of objects in Euclidean space. We formulate the embedding problem as a
joint ranking problem over a set of triplets, where each triplet captures the
relative similarities between three objects in the set. By exploiting recent
advances in robust ranking, t-ETE produces high-quality embeddings even in the
presence of a significant amount of noise and better preserves local scale than
known methods, such as t-STE and t-SNE. In particular, our method produces
significantly better results than t-SNE on signature datasets while also being
faster to compute.
| no_new_dataset | 0.952794 |
1703.07076 | Esben Jannik Bjerrum | Esben Jannik Bjerrum | SMILES Enumeration as Data Augmentation for Neural Network Modeling of
Molecules | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Simplified Molecular Input Line Entry System (SMILES) is a single line text
representation of a unique molecule. One molecule can however have multiple
SMILES strings, which is a reason that canonical SMILES have been defined,
which ensures a one to one correspondence between SMILES string and molecule.
Here the fact that multiple SMILES represent the same molecule is explored as a
technique for data augmentation of a molecular QSAR dataset modeled by a long
short term memory (LSTM) cell based neural network. The augmented dataset was
130 times bigger than the original. The network trained with the augmented
dataset shows better performance on a test set when compared to a model built
with only one canonical SMILES string per molecule. The correlation coefficient
R2 on the test set was improved from 0.56 to 0.66 when using SMILES
enumeration, and the root mean square error (RMS) likewise fell from 0.62 to
0.55. The technique also works in the prediction phase. By taking the average
per molecule of the predictions for the enumerated SMILES a further improvement
to a correlation coefficient of 0.68 and a RMS of 0.52 was found.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 07:13:13 GMT"
},
{
"version": "v2",
"created": "Wed, 17 May 2017 11:24:43 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Bjerrum",
"Esben Jannik",
""
]
] | TITLE: SMILES Enumeration as Data Augmentation for Neural Network Modeling of
Molecules
ABSTRACT: Simplified Molecular Input Line Entry System (SMILES) is a single line text
representation of a unique molecule. One molecule can however have multiple
SMILES strings, which is a reason that canonical SMILES have been defined,
which ensures a one to one correspondence between SMILES string and molecule.
Here the fact that multiple SMILES represent the same molecule is explored as a
technique for data augmentation of a molecular QSAR dataset modeled by a long
short term memory (LSTM) cell based neural network. The augmented dataset was
130 times bigger than the original. The network trained with the augmented
dataset shows better performance on a test set when compared to a model built
with only one canonical SMILES string per molecule. The correlation coefficient
R2 on the test set was improved from 0.56 to 0.66 when using SMILES
enumeration, and the root mean square error (RMS) likewise fell from 0.62 to
0.55. The technique also works in the prediction phase. By taking the average
per molecule of the predictions for the enumerated SMILES a further improvement
to a correlation coefficient of 0.68 and a RMS of 0.52 was found.
| no_new_dataset | 0.952706 |
1704.02003 | Samuel Pollard | Samuel Pollard and Boyana Norris | A Comparison of Parallel Graph Processing Implementations | 10 pages, 10 figures, Submitted to EuroPar 2017 and rejected. Revised
and submitted to IEEE Cluster 2017 | null | null | null | cs.PF cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapidly growing number of large network analysis problems has led to the
emergence of many parallel and distributed graph processing systems---one
survey in 2014 identified over 80. Since then, the landscape has evolved; some
packages have become inactive while more are being developed. Determining the
best approach for a given problem is infeasible for most developers. To enable
easy, rigorous, and repeatable comparison of the capabilities of such systems,
we present an approach and associated software for analyzing the performance
and scalability of parallel, open-source graph libraries. We demonstrate our
approach on five graph processing packages: GraphMat, the Graph500, the Graph
Algorithm Platform Benchmark Suite, GraphBIG, and PowerGraph using synthetic
and real-world datasets. We examine previously overlooked aspects of parallel
graph processing performance such as phases of execution and energy usage for
three algorithms: breadth first search, single source shortest paths, and
PageRank and compare our results to Graphalytics.
| [
{
"version": "v1",
"created": "Thu, 6 Apr 2017 19:48:37 GMT"
},
{
"version": "v2",
"created": "Wed, 17 May 2017 01:44:23 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Pollard",
"Samuel",
""
],
[
"Norris",
"Boyana",
""
]
] | TITLE: A Comparison of Parallel Graph Processing Implementations
ABSTRACT: The rapidly growing number of large network analysis problems has led to the
emergence of many parallel and distributed graph processing systems---one
survey in 2014 identified over 80. Since then, the landscape has evolved; some
packages have become inactive while more are being developed. Determining the
best approach for a given problem is infeasible for most developers. To enable
easy, rigorous, and repeatable comparison of the capabilities of such systems,
we present an approach and associated software for analyzing the performance
and scalability of parallel, open-source graph libraries. We demonstrate our
approach on five graph processing packages: GraphMat, the Graph500, the Graph
Algorithm Platform Benchmark Suite, GraphBIG, and PowerGraph using synthetic
and real-world datasets. We examine previously overlooked aspects of parallel
graph processing performance such as phases of execution and energy usage for
three algorithms: breadth first search, single source shortest paths, and
PageRank and compare our results to Graphalytics.
| no_new_dataset | 0.941385 |
1705.04612 | Esben Jannik Bjerrum | Esben Jannik Bjerrum, Richard Threlfall | Molecular Generation with Recurrent Neural Networks (RNNs) | null | null | null | null | cs.LG q-bio.BM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The potential number of drug like small molecules is estimated to be between
10^23 and 10^60 while current databases of known compounds are orders of
magnitude smaller with approximately 10^8 compounds. This discrepancy has led
to an interest in generating virtual libraries using hand crafted chemical
rules and fragment based methods to cover a larger area of chemical space and
generate chemical libraries for use in in silico drug discovery endeavors. Here
it is explored to what extent a recurrent neural network with long short term
memory cells can figure out sensible chemical rules and generate synthesizable
molecules by being trained on existing compounds encoded as SMILES. The
networks can to a high extent generate novel, but chemically sensible
molecules. The properties of the molecules are tuned by training on two
different datasets consisting of fragment like molecules and drug like
molecules. The produced molecules and the training databases have very similar
distributions of molar weight, predicted logP, number of hydrogen bond
acceptors and donors, number of rotatable bonds and topological polar surface
area when compared to their respective training sets. The compounds are for the
most cases synthesizable as assessed with SA score and Wiley ChemPlanner.
| [
{
"version": "v1",
"created": "Fri, 12 May 2017 14:56:09 GMT"
},
{
"version": "v2",
"created": "Wed, 17 May 2017 10:55:22 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Bjerrum",
"Esben Jannik",
""
],
[
"Threlfall",
"Richard",
""
]
] | TITLE: Molecular Generation with Recurrent Neural Networks (RNNs)
ABSTRACT: The potential number of drug like small molecules is estimated to be between
10^23 and 10^60 while current databases of known compounds are orders of
magnitude smaller with approximately 10^8 compounds. This discrepancy has led
to an interest in generating virtual libraries using hand crafted chemical
rules and fragment based methods to cover a larger area of chemical space and
generate chemical libraries for use in in silico drug discovery endeavors. Here
it is explored to what extent a recurrent neural network with long short term
memory cells can figure out sensible chemical rules and generate synthesizable
molecules by being trained on existing compounds encoded as SMILES. The
networks can to a high extent generate novel, but chemically sensible
molecules. The properties of the molecules are tuned by training on two
different datasets consisting of fragment like molecules and drug like
molecules. The produced molecules and the training databases have very similar
distributions of molar weight, predicted logP, number of hydrogen bond
acceptors and donors, number of rotatable bonds and topological polar surface
area when compared to their respective training sets. The compounds are for the
most cases synthesizable as assessed with SA score and Wiley ChemPlanner.
| no_new_dataset | 0.952264 |
1705.05986 | Srinivasan Parthasarathy | Yanjie Fu, Charu Aggarwal, Srinivasan Parthasarathy, Deepak S. Turaga,
Hui Xiong | REMIX: Automated Exploration for Interactive Outlier Detection | To appear in KDD 2017 | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Outlier detection is the identification of points in a dataset that do not
conform to the norm. Outlier detection is highly sensitive to the choice of the
detection algorithm and the feature subspace used by the algorithm. Extracting
domain-relevant insights from outliers needs systematic exploration of these
choices since diverse outlier sets could lead to complementary insights. This
challenge is especially acute in an interactive setting, where the choices must
be explored in a time-constrained manner. In this work, we present REMIX, the
first system to address the problem of outlier detection in an interactive
setting. REMIX uses a novel mixed integer programming (MIP) formulation for
automatically selecting and executing a diverse set of outlier detectors within
a time limit. This formulation incorporates multiple aspects such as (i) an
upper limit on the total execution time of detectors (ii) diversity in the
space of algorithms and features, and (iii) meta-learning for evaluating the
cost and utility of detectors. REMIX provides two distinct ways for the analyst
to consume its results: (i) a partitioning of the detectors explored by REMIX
into perspectives through low-rank non-negative matrix factorization; each
perspective can be easily visualized as an intuitive heatmap of experiments
versus outliers, and (ii) an ensembled set of outliers which combines outlier
scores from all detectors. We demonstrate the benefits of REMIX through
extensive empirical validation on real-world data.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 02:17:48 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Fu",
"Yanjie",
""
],
[
"Aggarwal",
"Charu",
""
],
[
"Parthasarathy",
"Srinivasan",
""
],
[
"Turaga",
"Deepak S.",
""
],
[
"Xiong",
"Hui",
""
]
] | TITLE: REMIX: Automated Exploration for Interactive Outlier Detection
ABSTRACT: Outlier detection is the identification of points in a dataset that do not
conform to the norm. Outlier detection is highly sensitive to the choice of the
detection algorithm and the feature subspace used by the algorithm. Extracting
domain-relevant insights from outliers needs systematic exploration of these
choices since diverse outlier sets could lead to complementary insights. This
challenge is especially acute in an interactive setting, where the choices must
be explored in a time-constrained manner. In this work, we present REMIX, the
first system to address the problem of outlier detection in an interactive
setting. REMIX uses a novel mixed integer programming (MIP) formulation for
automatically selecting and executing a diverse set of outlier detectors within
a time limit. This formulation incorporates multiple aspects such as (i) an
upper limit on the total execution time of detectors (ii) diversity in the
space of algorithms and features, and (iii) meta-learning for evaluating the
cost and utility of detectors. REMIX provides two distinct ways for the analyst
to consume its results: (i) a partitioning of the detectors explored by REMIX
into perspectives through low-rank non-negative matrix factorization; each
perspective can be easily visualized as an intuitive heatmap of experiments
versus outliers, and (ii) an ensembled set of outliers which combines outlier
scores from all detectors. We demonstrate the benefits of REMIX through
extensive empirical validation on real-world data.
| no_new_dataset | 0.945651 |
1705.05992 | Jun Zhang | Xu Tian, Jun Zhang, Zejun Ma, Yi He, Juan Wei | Frame Stacking and Retaining for Recurrent Neural Network Acoustic Model | 5 pages | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Frame stacking is broadly applied in end-to-end neural network training like
connectionist temporal classification (CTC), and it leads to more accurate
models and faster decoding. However, it is not well-suited to conventional
neural network based on context-dependent state acoustic model, if the decoder
is unchanged. In this paper, we propose a novel frame retaining method which is
applied in decoding. The system which combined frame retaining with frame
stacking could reduces the time consumption of both training and decoding. Long
short-term memory (LSTM) recurrent neural networks (RNNs) using it achieve
almost linear training speedup and reduces relative 41\% real time factor
(RTF). At the same time, recognition performance is no degradation or improves
sightly on Shenma voice search dataset in Mandarin.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 02:34:27 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Tian",
"Xu",
""
],
[
"Zhang",
"Jun",
""
],
[
"Ma",
"Zejun",
""
],
[
"He",
"Yi",
""
],
[
"Wei",
"Juan",
""
]
] | TITLE: Frame Stacking and Retaining for Recurrent Neural Network Acoustic Model
ABSTRACT: Frame stacking is broadly applied in end-to-end neural network training like
connectionist temporal classification (CTC), and it leads to more accurate
models and faster decoding. However, it is not well-suited to conventional
neural network based on context-dependent state acoustic model, if the decoder
is unchanged. In this paper, we propose a novel frame retaining method which is
applied in decoding. The system which combined frame retaining with frame
stacking could reduces the time consumption of both training and decoding. Long
short-term memory (LSTM) recurrent neural networks (RNNs) using it achieve
almost linear training speedup and reduces relative 41\% real time factor
(RTF). At the same time, recognition performance is no degradation or improves
sightly on Shenma voice search dataset in Mandarin.
| no_new_dataset | 0.951997 |
1705.05998 | Tao Xiong | Dong Yang, Tao Xiong, Daguang Xu, Qiangui Huang, David Liu, S.Kevin
Zhou, Zhoubing Xu, JinHyeong Park, Mingqing Chen, Trac D. Tran, Sang Peter
Chin, Dimitris Metaxas, Dorin Comaniciu | Automatic Vertebra Labeling in Large-Scale 3D CT using Deep
Image-to-Image Network with Message Passing and Sparsity Regularization | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic localization and labeling of vertebra in 3D medical images plays an
important role in many clinical tasks, including pathological diagnosis,
surgical planning and postoperative assessment. However, the unusual conditions
of pathological cases, such as the abnormal spine curvature, bright visual
imaging artifacts caused by metal implants, and the limited field of view,
increase the difficulties of accurate localization. In this paper, we propose
an automatic and fast algorithm to localize and label the vertebra centroids in
3D CT volumes. First, we deploy a deep image-to-image network (DI2IN) to
initialize vertebra locations, employing the convolutional encoder-decoder
architecture together with multi-level feature concatenation and deep
supervision. Next, the centroid probability maps from DI2IN are iteratively
evolved with the message passing schemes based on the mutual relation of
vertebra centroids. Finally, the localization results are refined with sparsity
regularization. The proposed method is evaluated on a public dataset of 302
spine CT volumes with various pathologies. Our method outperforms other
state-of-the-art methods in terms of localization accuracy. The run time is
around 3 seconds on average per case. To further boost the performance, we
retrain the DI2IN on additional 1000+ 3D CT volumes from different patients. To
the best of our knowledge, this is the first time more than 1000 3D CT volumes
with expert annotation are adopted in experiments for the anatomic landmark
detection tasks. Our experimental results show that training with such a large
dataset significantly improves the performance and the overall identification
rate, for the first time by our knowledge, reaches 90 %.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 03:56:14 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Yang",
"Dong",
""
],
[
"Xiong",
"Tao",
""
],
[
"Xu",
"Daguang",
""
],
[
"Huang",
"Qiangui",
""
],
[
"Liu",
"David",
""
],
[
"Zhou",
"S. Kevin",
""
],
[
"Xu",
"Zhoubing",
""
],
[
"Park",
"JinHyeong",
""
],
[
"Chen",
"Mingqing",
""
],
[
"Tran",
"Trac D.",
""
],
[
"Chin",
"Sang Peter",
""
],
[
"Metaxas",
"Dimitris",
""
],
[
"Comaniciu",
"Dorin",
""
]
] | TITLE: Automatic Vertebra Labeling in Large-Scale 3D CT using Deep
Image-to-Image Network with Message Passing and Sparsity Regularization
ABSTRACT: Automatic localization and labeling of vertebra in 3D medical images plays an
important role in many clinical tasks, including pathological diagnosis,
surgical planning and postoperative assessment. However, the unusual conditions
of pathological cases, such as the abnormal spine curvature, bright visual
imaging artifacts caused by metal implants, and the limited field of view,
increase the difficulties of accurate localization. In this paper, we propose
an automatic and fast algorithm to localize and label the vertebra centroids in
3D CT volumes. First, we deploy a deep image-to-image network (DI2IN) to
initialize vertebra locations, employing the convolutional encoder-decoder
architecture together with multi-level feature concatenation and deep
supervision. Next, the centroid probability maps from DI2IN are iteratively
evolved with the message passing schemes based on the mutual relation of
vertebra centroids. Finally, the localization results are refined with sparsity
regularization. The proposed method is evaluated on a public dataset of 302
spine CT volumes with various pathologies. Our method outperforms other
state-of-the-art methods in terms of localization accuracy. The run time is
around 3 seconds on average per case. To further boost the performance, we
retrain the DI2IN on additional 1000+ 3D CT volumes from different patients. To
the best of our knowledge, this is the first time more than 1000 3D CT volumes
with expert annotation are adopted in experiments for the anatomic landmark
detection tasks. Our experimental results show that training with such a large
dataset significantly improves the performance and the overall identification
rate, for the first time by our knowledge, reaches 90 %.
| no_new_dataset | 0.950824 |
1705.06000 | Abhishek Sharma | Abhishek Sharma | One Shot Joint Colocalization and Cosegmentation | 8 pages, Under Review | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel framework in which image cosegmentation and
colocalization are cast into a single optimization problem that integrates
information from low level appearance cues with that of high level localization
cues in a very weakly supervised manner. In contrast to multi-task learning
paradigm that learns similar tasks using a shared representation, the proposed
framework leverages two representations at different levels and simultaneously
discriminates between foreground and background at the bounding box and
superpixel level using discriminative clustering. We show empirically that
constraining the two problems at different scales enables the transfer of
semantic localization cues to improve cosegmentation output whereas local
appearance based segmentation cues help colocalization. The unified framework
outperforms strong baseline approaches, of learning the two problems
separately, by a large margin on four benchmark datasets. Furthermore, it
obtains competitive results compared to the state of the art for cosegmentation
on two benchmark datasets and second best result for colocalization on Pascal
VOC 2007.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 04:18:19 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Sharma",
"Abhishek",
""
]
] | TITLE: One Shot Joint Colocalization and Cosegmentation
ABSTRACT: This paper presents a novel framework in which image cosegmentation and
colocalization are cast into a single optimization problem that integrates
information from low level appearance cues with that of high level localization
cues in a very weakly supervised manner. In contrast to multi-task learning
paradigm that learns similar tasks using a shared representation, the proposed
framework leverages two representations at different levels and simultaneously
discriminates between foreground and background at the bounding box and
superpixel level using discriminative clustering. We show empirically that
constraining the two problems at different scales enables the transfer of
semantic localization cues to improve cosegmentation output whereas local
appearance based segmentation cues help colocalization. The unified framework
outperforms strong baseline approaches, of learning the two problems
separately, by a large margin on four benchmark datasets. Furthermore, it
obtains competitive results compared to the state of the art for cosegmentation
on two benchmark datasets and second best result for colocalization on Pascal
VOC 2007.
| no_new_dataset | 0.950641 |
1705.06057 | Nicolas Audebert | Nicolas Audebert (Palaiseau, OBELIX), Bertrand Le Saux (Palaiseau),
S\'ebastien Lef\`evre (OBELIX) | Joint Learning from Earth Observation and OpenStreetMap Data to Get
Faster Better Semantic Maps | null | EARTHVISION 2017 IEEE/ISPRS CVPR Workshop. Large Scale Computer
Vision for Remote Sensing Imagery, Jul 2017, Honolulu, United States. 2017 | null | null | cs.CV cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we investigate the use of OpenStreetMap data for semantic
labeling of Earth Observation images. Deep neural networks have been used in
the past for remote sensing data classification from various sensors, including
multispectral, hyperspectral, SAR and LiDAR data. While OpenStreetMap has
already been used as ground truth data for training such networks, this
abundant data source remains rarely exploited as an input information layer. In
this paper, we study different use cases and deep network architectures to
leverage OpenStreetMap data for semantic labeling of aerial and satellite
images. Especially , we look into fusion based architectures and coarse-to-fine
segmentation to include the OpenStreetMap layer into multispectral-based deep
fully convolutional networks. We illustrate how these methods can be
successfully used on two public datasets: ISPRS Potsdam and DFC2017. We show
that OpenStreetMap data can efficiently be integrated into the vision-based
deep learning models and that it significantly improves both the accuracy
performance and the convergence speed of the networks.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 09:07:08 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Audebert",
"Nicolas",
"",
"Palaiseau, OBELIX"
],
[
"Saux",
"Bertrand Le",
"",
"Palaiseau"
],
[
"Lefèvre",
"Sébastien",
"",
"OBELIX"
]
] | TITLE: Joint Learning from Earth Observation and OpenStreetMap Data to Get
Faster Better Semantic Maps
ABSTRACT: In this work, we investigate the use of OpenStreetMap data for semantic
labeling of Earth Observation images. Deep neural networks have been used in
the past for remote sensing data classification from various sensors, including
multispectral, hyperspectral, SAR and LiDAR data. While OpenStreetMap has
already been used as ground truth data for training such networks, this
abundant data source remains rarely exploited as an input information layer. In
this paper, we study different use cases and deep network architectures to
leverage OpenStreetMap data for semantic labeling of aerial and satellite
images. Especially , we look into fusion based architectures and coarse-to-fine
segmentation to include the OpenStreetMap layer into multispectral-based deep
fully convolutional networks. We illustrate how these methods can be
successfully used on two public datasets: ISPRS Potsdam and DFC2017. We show
that OpenStreetMap data can efficiently be integrated into the vision-based
deep learning models and that it significantly improves both the accuracy
performance and the convergence speed of the networks.
| no_new_dataset | 0.952794 |
1705.06201 | Anirudh Vemula | Anirudh Vemula, Katharina Muelling and Jean Oh | Modeling Cooperative Navigation in Dense Human Crowds | Accepted at ICRA 2017 | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For robots to be a part of our daily life, they need to be able to navigate
among crowds not only safely but also in a socially compliant fashion. This is
a challenging problem because humans tend to navigate by implicitly cooperating
with one another to avoid collisions, while heading toward their respective
destinations. Previous approaches have used hand-crafted functions based on
proximity to model human-human and human-robot interactions. However, these
approaches can only model simple interactions and fail to generalize for
complex crowded settings. In this paper, we develop an approach that models the
joint distribution over future trajectories of all interacting agents in the
crowd, through a local interaction model that we train using real human
trajectory data. The interaction model infers the velocity of each agent based
on the spatial orientation of other agents in his vicinity. During prediction,
our approach infers the goal of the agent from its past trajectory and uses the
learned model to predict its future trajectory. We demonstrate the performance
of our method against a state-of-the-art approach on a public dataset and show
that our model outperforms when predicting future trajectories for longer
horizons.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 15:12:46 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Vemula",
"Anirudh",
""
],
[
"Muelling",
"Katharina",
""
],
[
"Oh",
"Jean",
""
]
] | TITLE: Modeling Cooperative Navigation in Dense Human Crowds
ABSTRACT: For robots to be a part of our daily life, they need to be able to navigate
among crowds not only safely but also in a socially compliant fashion. This is
a challenging problem because humans tend to navigate by implicitly cooperating
with one another to avoid collisions, while heading toward their respective
destinations. Previous approaches have used hand-crafted functions based on
proximity to model human-human and human-robot interactions. However, these
approaches can only model simple interactions and fail to generalize for
complex crowded settings. In this paper, we develop an approach that models the
joint distribution over future trajectories of all interacting agents in the
crowd, through a local interaction model that we train using real human
trajectory data. The interaction model infers the velocity of each agent based
on the spatial orientation of other agents in his vicinity. During prediction,
our approach infers the goal of the agent from its past trajectory and uses the
learned model to predict its future trajectory. We demonstrate the performance
of our method against a state-of-the-art approach on a public dataset and show
that our model outperforms when predicting future trajectories for longer
horizons.
| no_new_dataset | 0.94428 |
1705.06273 | Franck Dernoncourt | Ji Young Lee, Franck Dernoncourt, Peter Szolovits | Transfer Learning for Named-Entity Recognition with Neural Networks | The first two authors contributed equally to this work | null | null | null | cs.CL cs.AI cs.NE stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent approaches based on artificial neural networks (ANNs) have shown
promising results for named-entity recognition (NER). In order to achieve high
performances, ANNs need to be trained on a large labeled dataset. However,
labels might be difficult to obtain for the dataset on which the user wants to
perform NER: label scarcity is particularly pronounced for patient note
de-identification, which is an instance of NER. In this work, we analyze to
what extent transfer learning may address this issue. In particular, we
demonstrate that transferring an ANN model trained on a large labeled dataset
to another dataset with a limited number of labels improves upon the
state-of-the-art results on two different datasets for patient note
de-identification.
| [
{
"version": "v1",
"created": "Wed, 17 May 2017 17:45:15 GMT"
}
] | 2017-05-18T00:00:00 | [
[
"Lee",
"Ji Young",
""
],
[
"Dernoncourt",
"Franck",
""
],
[
"Szolovits",
"Peter",
""
]
] | TITLE: Transfer Learning for Named-Entity Recognition with Neural Networks
ABSTRACT: Recent approaches based on artificial neural networks (ANNs) have shown
promising results for named-entity recognition (NER). In order to achieve high
performances, ANNs need to be trained on a large labeled dataset. However,
labels might be difficult to obtain for the dataset on which the user wants to
perform NER: label scarcity is particularly pronounced for patient note
de-identification, which is an instance of NER. In this work, we analyze to
what extent transfer learning may address this issue. In particular, we
demonstrate that transferring an ANN model trained on a large labeled dataset
to another dataset with a limited number of labels improves upon the
state-of-the-art results on two different datasets for patient note
de-identification.
| no_new_dataset | 0.958304 |
1607.04573 | Luiz Gustavo Hafemann | Luiz G. Hafemann, Robert Sabourin, Luiz S. Oliveira | Analyzing features learned for Offline Signature Verification using Deep
CNNs | Accepted as a conference paper to ICPR 2016 | null | 10.1109/ICPR.2016.7900092 | null | cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Research on Offline Handwritten Signature Verification explored a large
variety of handcrafted feature extractors, ranging from graphology, texture
descriptors to interest points. In spite of advancements in the last decades,
performance of such systems is still far from optimal when we test the systems
against skilled forgeries - signature forgeries that target a particular
individual. In previous research, we proposed a formulation of the problem to
learn features from data (signature images) in a Writer-Independent format,
using Deep Convolutional Neural Networks (CNNs), seeking to improve performance
on the task. In this research, we push further the performance of such method,
exploring a range of architectures, and obtaining a large improvement in
state-of-the-art performance on the GPDS dataset, the largest publicly
available dataset on the task. In the GPDS-160 dataset, we obtained an Equal
Error Rate of 2.74%, compared to 6.97% in the best result published in
literature (that used a combination of multiple classifiers). We also present a
visual analysis of the feature space learned by the model, and an analysis of
the errors made by the classifier. Our analysis shows that the model is very
effective in separating signatures that have a different global appearance,
while being particularly vulnerable to forgeries that very closely resemble
genuine signatures, even if their line quality is bad, which is the case of
slowly-traced forgeries.
| [
{
"version": "v1",
"created": "Fri, 15 Jul 2016 16:35:20 GMT"
},
{
"version": "v2",
"created": "Fri, 26 Aug 2016 14:52:55 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Hafemann",
"Luiz G.",
""
],
[
"Sabourin",
"Robert",
""
],
[
"Oliveira",
"Luiz S.",
""
]
] | TITLE: Analyzing features learned for Offline Signature Verification using Deep
CNNs
ABSTRACT: Research on Offline Handwritten Signature Verification explored a large
variety of handcrafted feature extractors, ranging from graphology, texture
descriptors to interest points. In spite of advancements in the last decades,
performance of such systems is still far from optimal when we test the systems
against skilled forgeries - signature forgeries that target a particular
individual. In previous research, we proposed a formulation of the problem to
learn features from data (signature images) in a Writer-Independent format,
using Deep Convolutional Neural Networks (CNNs), seeking to improve performance
on the task. In this research, we push further the performance of such method,
exploring a range of architectures, and obtaining a large improvement in
state-of-the-art performance on the GPDS dataset, the largest publicly
available dataset on the task. In the GPDS-160 dataset, we obtained an Equal
Error Rate of 2.74%, compared to 6.97% in the best result published in
literature (that used a combination of multiple classifiers). We also present a
visual analysis of the feature space learned by the model, and an analysis of
the errors made by the classifier. Our analysis shows that the model is very
effective in separating signatures that have a different global appearance,
while being particularly vulnerable to forgeries that very closely resemble
genuine signatures, even if their line quality is bad, which is the case of
slowly-traced forgeries.
| no_new_dataset | 0.94474 |
1612.00558 | Basura Fernando | Basura Fernando, Sareh Shirazi and Stephen Gould | Unsupervised Human Action Detection by Action Matching | IEEE International Conference on Computer Vision and Pattern
Recognition CVPR 2017 Workshops | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a new task of unsupervised action detection by action matching.
Given two long videos, the objective is to temporally detect all pairs of
matching video segments. A pair of video segments are matched if they share the
same human action. The task is category independent---it does not matter what
action is being performed---and no supervision is used to discover such video
segments. Unsupervised action detection by action matching allows us to align
videos in a meaningful manner. As such, it can be used to discover new action
categories or as an action proposal technique within, say, an action detection
pipeline. Moreover, it is a useful pre-processing step for generating video
highlights, e.g., from sports videos.
We present an effective and efficient method for unsupervised action
detection. We use an unsupervised temporal encoding method and exploit the
temporal consistency in human actions to obtain candidate action segments. We
evaluate our method on this challenging task using three activity recognition
benchmarks, namely, the MPII Cooking activities dataset, the THUMOS15 action
detection benchmark and a new dataset called the IKEA dataset. On the MPII
Cooking dataset we detect action segments with a precision of 21.6% and recall
of 11.7% over 946 long video pairs and over 5000 ground truth action segments.
Similarly, on THUMOS dataset we obtain 18.4% precision and 25.1% recall over
5094 ground truth action segment pairs.
| [
{
"version": "v1",
"created": "Fri, 2 Dec 2016 03:39:38 GMT"
},
{
"version": "v2",
"created": "Mon, 3 Apr 2017 03:36:17 GMT"
},
{
"version": "v3",
"created": "Wed, 5 Apr 2017 06:18:22 GMT"
},
{
"version": "v4",
"created": "Tue, 16 May 2017 00:56:24 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Fernando",
"Basura",
""
],
[
"Shirazi",
"Sareh",
""
],
[
"Gould",
"Stephen",
""
]
] | TITLE: Unsupervised Human Action Detection by Action Matching
ABSTRACT: We propose a new task of unsupervised action detection by action matching.
Given two long videos, the objective is to temporally detect all pairs of
matching video segments. A pair of video segments are matched if they share the
same human action. The task is category independent---it does not matter what
action is being performed---and no supervision is used to discover such video
segments. Unsupervised action detection by action matching allows us to align
videos in a meaningful manner. As such, it can be used to discover new action
categories or as an action proposal technique within, say, an action detection
pipeline. Moreover, it is a useful pre-processing step for generating video
highlights, e.g., from sports videos.
We present an effective and efficient method for unsupervised action
detection. We use an unsupervised temporal encoding method and exploit the
temporal consistency in human actions to obtain candidate action segments. We
evaluate our method on this challenging task using three activity recognition
benchmarks, namely, the MPII Cooking activities dataset, the THUMOS15 action
detection benchmark and a new dataset called the IKEA dataset. On the MPII
Cooking dataset we detect action segments with a precision of 21.6% and recall
of 11.7% over 946 long video pairs and over 5000 ground truth action segments.
Similarly, on THUMOS dataset we obtain 18.4% precision and 25.1% recall over
5094 ground truth action segment pairs.
| new_dataset | 0.96225 |
1704.01508 | Zafar Gilani | Zafar Gilani, Reza Farahbakhsh, Gareth Tyson, Liang Wang, Jon
Crowcroft | An in-depth characterisation of Bots and Humans on Twitter | This is a technical report of 18 pages including references | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent research has shown a substantial active presence of bots in online
social networks (OSNs). In this paper we utilise our past work on studying bots
(Stweeler) to comparatively analyse the usage and impact of bots and humans on
Twitter, one of the largest OSNs in the world. We collect a large-scale Twitter
dataset and define various metrics based on tweet metadata. We divide and
filter the dataset in four popularity groups in terms of number of followers.
Using a human annotation task we assign 'bot' and 'human' ground-truth labels
to the dataset, and compare the annotations against an online bot detection
tool for evaluation. We then ask a series of questions to discern important
behavioural bot and human characteristics using metrics within and among four
popularity groups. From the comparative analysis we draw important differences
as well as surprising similarities between the two entities, thus paving the
way for reliable classification of automated political infiltration,
advertisement campaigns, and general bot detection.
| [
{
"version": "v1",
"created": "Wed, 5 Apr 2017 16:17:41 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Gilani",
"Zafar",
""
],
[
"Farahbakhsh",
"Reza",
""
],
[
"Tyson",
"Gareth",
""
],
[
"Wang",
"Liang",
""
],
[
"Crowcroft",
"Jon",
""
]
] | TITLE: An in-depth characterisation of Bots and Humans on Twitter
ABSTRACT: Recent research has shown a substantial active presence of bots in online
social networks (OSNs). In this paper we utilise our past work on studying bots
(Stweeler) to comparatively analyse the usage and impact of bots and humans on
Twitter, one of the largest OSNs in the world. We collect a large-scale Twitter
dataset and define various metrics based on tweet metadata. We divide and
filter the dataset in four popularity groups in terms of number of followers.
Using a human annotation task we assign 'bot' and 'human' ground-truth labels
to the dataset, and compare the annotations against an online bot detection
tool for evaluation. We then ask a series of questions to discern important
behavioural bot and human characteristics using metrics within and among four
popularity groups. From the comparative analysis we draw important differences
as well as surprising similarities between the two entities, thus paving the
way for reliable classification of automated political infiltration,
advertisement campaigns, and general bot detection.
| no_new_dataset | 0.946547 |
1705.04353 | Daniel Larremore | Andrew Berdahl, Uttam Bhat, Vanessa Ferdinand, Joshua Garland, Keyan
Ghazi-Zahedi, Justin Grana, Joshua A. Grochow, Elizabeth Hobson, Yoav Kallus,
Christopher P. Kempes, Artemy Kolchinsky, Daniel B. Larremore, Eric Libby,
Eleanor A. Power, and Brendan D. Tracey (Santa Fe Institute Postdocs) | On the records | This paper was produced, from conception of idea, to execution, to
writing, by a team in just 72 hours (see Appendix) | null | null | null | physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | World record setting has long attracted public interest and scientific
investigation. Extremal records summarize the limits of the space explored by a
process, and the historical progression of a record sheds light on the
underlying dynamics of the process. Existing analyses of prediction,
statistical properties, and ultimate limits of record progressions have focused
on particular domains. However, a broad perspective on how record progressions
vary across different spheres of activity needs further development. Here we
employ cross-cutting metrics to compare records across a variety of domains,
including sports, games, biological evolution, and technological development.
We find that these domains exhibit characteristic statistical signatures in
terms of rates of improvement, "burstiness" of record-breaking time series, and
the acceleration of the record breaking process. Specifically, sports and games
exhibit the slowest rate of improvement and a wide range of rates of
"burstiness." Technology improves at a much faster rate and, unlike other
domains, tends to show acceleration in records. Many biological and
technological processes are characterized by constant rates of improvement,
showing less burstiness than sports and games. It is important to understand
how these statistical properties of record progression emerge from the
underlying dynamics. Towards this end, we conduct a detailed analysis of a
particular record-setting event: elite marathon running. In this domain, we
find that studying record-setting data alone can obscure many of the structural
properties of the underlying process. The marathon study also illustrates how
some of the standard statistical assumptions underlying record progression
models may be inappropriate or commonly violated in real-world datasets.
| [
{
"version": "v1",
"created": "Thu, 11 May 2017 18:59:43 GMT"
},
{
"version": "v2",
"created": "Mon, 15 May 2017 18:00:16 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Berdahl",
"Andrew",
"",
"Santa Fe Institute Postdocs"
],
[
"Bhat",
"Uttam",
"",
"Santa Fe Institute Postdocs"
],
[
"Ferdinand",
"Vanessa",
"",
"Santa Fe Institute Postdocs"
],
[
"Garland",
"Joshua",
"",
"Santa Fe Institute Postdocs"
],
[
"Ghazi-Zahedi",
"Keyan",
"",
"Santa Fe Institute Postdocs"
],
[
"Grana",
"Justin",
"",
"Santa Fe Institute Postdocs"
],
[
"Grochow",
"Joshua A.",
"",
"Santa Fe Institute Postdocs"
],
[
"Hobson",
"Elizabeth",
"",
"Santa Fe Institute Postdocs"
],
[
"Kallus",
"Yoav",
"",
"Santa Fe Institute Postdocs"
],
[
"Kempes",
"Christopher P.",
"",
"Santa Fe Institute Postdocs"
],
[
"Kolchinsky",
"Artemy",
"",
"Santa Fe Institute Postdocs"
],
[
"Larremore",
"Daniel B.",
"",
"Santa Fe Institute Postdocs"
],
[
"Libby",
"Eric",
"",
"Santa Fe Institute Postdocs"
],
[
"Power",
"Eleanor A.",
"",
"Santa Fe Institute Postdocs"
],
[
"Tracey",
"Brendan D.",
"",
"Santa Fe Institute Postdocs"
]
] | TITLE: On the records
ABSTRACT: World record setting has long attracted public interest and scientific
investigation. Extremal records summarize the limits of the space explored by a
process, and the historical progression of a record sheds light on the
underlying dynamics of the process. Existing analyses of prediction,
statistical properties, and ultimate limits of record progressions have focused
on particular domains. However, a broad perspective on how record progressions
vary across different spheres of activity needs further development. Here we
employ cross-cutting metrics to compare records across a variety of domains,
including sports, games, biological evolution, and technological development.
We find that these domains exhibit characteristic statistical signatures in
terms of rates of improvement, "burstiness" of record-breaking time series, and
the acceleration of the record breaking process. Specifically, sports and games
exhibit the slowest rate of improvement and a wide range of rates of
"burstiness." Technology improves at a much faster rate and, unlike other
domains, tends to show acceleration in records. Many biological and
technological processes are characterized by constant rates of improvement,
showing less burstiness than sports and games. It is important to understand
how these statistical properties of record progression emerge from the
underlying dynamics. Towards this end, we conduct a detailed analysis of a
particular record-setting event: elite marathon running. In this domain, we
find that studying record-setting data alone can obscure many of the structural
properties of the underlying process. The marathon study also illustrates how
some of the standard statistical assumptions underlying record progression
models may be inappropriate or commonly violated in real-world datasets.
| no_new_dataset | 0.928862 |
1705.05219 | Sobhan Moosavi | Sobhan Moosavi, Behrooz Omidvar-Tehrani, R. Bruce Craig, Rajiv Ramnath | Annotation of Car Trajectories based on Driving Patterns | A 10 pages technical report which described the process of preparing
a ground-truth dataset | null | null | null | cs.OH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nowadays, the ubiquity of various sensors enables the collection of
voluminous datasets of car trajectories. Such datasets enable analysts to make
sense of driving patterns and behaviors: in order to understand the behavior of
drivers, one approach is to break a trajectory into its underlying patterns and
then analyze that trajectory in terms of derived patterns. The process of
trajectory segmentation is a function of various resources including a set of
ground truth trajectories with their driving patterns. To the best of our
knowledge, no such ground-truth dataset exists in the literature. In this
paper, we describe a trajectory annotation framework and report our results to
annotate a dataset of personal car trajectories. Our annotation methodology
consists of a crowd-sourcing task followed by a precise process of aggregation.
Our annotation process consists of two granularity levels, one to specify the
annotation (segment border) and the other one to describe the type of the
segment (e.g. speed-up, turn, merge, etc.). The output of our project, Dataset
of Annotated Car Trajectories (DACT), is available online at
https://figshare.com/articles/dact_dataset_of_annotated_car_trajectories/5005289 .
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 13:30:36 GMT"
},
{
"version": "v2",
"created": "Tue, 16 May 2017 14:48:34 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Moosavi",
"Sobhan",
""
],
[
"Omidvar-Tehrani",
"Behrooz",
""
],
[
"Craig",
"R. Bruce",
""
],
[
"Ramnath",
"Rajiv",
""
]
] | TITLE: Annotation of Car Trajectories based on Driving Patterns
ABSTRACT: Nowadays, the ubiquity of various sensors enables the collection of
voluminous datasets of car trajectories. Such datasets enable analysts to make
sense of driving patterns and behaviors: in order to understand the behavior of
drivers, one approach is to break a trajectory into its underlying patterns and
then analyze that trajectory in terms of derived patterns. The process of
trajectory segmentation is a function of various resources including a set of
ground truth trajectories with their driving patterns. To the best of our
knowledge, no such ground-truth dataset exists in the literature. In this
paper, we describe a trajectory annotation framework and report our results to
annotate a dataset of personal car trajectories. Our annotation methodology
consists of a crowd-sourcing task followed by a precise process of aggregation.
Our annotation process consists of two granularity levels, one to specify the
annotation (segment border) and the other one to describe the type of the
segment (e.g. speed-up, turn, merge, etc.). The output of our project, Dataset
of Annotated Car Trajectories (DACT), is available online at
https://figshare.com/articles/dact_dataset_of_annotated_car_trajectories/5005289 .
| new_dataset | 0.968709 |
1705.05435 | Mehmet Turan | Mehmet Turan, Yasin Almalioglu, Ender Konukoglu, Metin Sitti | A Deep Learning Based 6 Degree-of-Freedom Localization Method for
Endoscopic Capsule Robots | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a robust deep learning based 6 degrees-of-freedom (DoF)
localization system for endoscopic capsule robots. Our system mainly focuses on
localization of endoscopic capsule robots inside the GI tract using only visual
information captured by a mono camera integrated to the robot. The proposed
system is a 23-layer deep convolutional neural network (CNN) that is capable to
estimate the pose of the robot in real time using a standard CPU. The dataset
for the evaluation of the system was recorded inside a surgical human stomach
model with realistic surface texture, softness, and surface liquid properties
so that the pre-trained CNN architecture can be transferred confidently into a
real endoscopic scenario. An average error of 7:1% and 3:4% for translation and
rotation has been obtained, respectively. The results accomplished from the
experiments demonstrate that a CNN pre-trained with raw 2D endoscopic images
performs accurately inside the GI tract and is robust to various challenges
posed by reflection distortions, lens imperfections, vignetting, noise, motion
blur, low resolution, and lack of unique landmarks to track.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 20:33:37 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Turan",
"Mehmet",
""
],
[
"Almalioglu",
"Yasin",
""
],
[
"Konukoglu",
"Ender",
""
],
[
"Sitti",
"Metin",
""
]
] | TITLE: A Deep Learning Based 6 Degree-of-Freedom Localization Method for
Endoscopic Capsule Robots
ABSTRACT: We present a robust deep learning based 6 degrees-of-freedom (DoF)
localization system for endoscopic capsule robots. Our system mainly focuses on
localization of endoscopic capsule robots inside the GI tract using only visual
information captured by a mono camera integrated to the robot. The proposed
system is a 23-layer deep convolutional neural network (CNN) that is capable to
estimate the pose of the robot in real time using a standard CPU. The dataset
for the evaluation of the system was recorded inside a surgical human stomach
model with realistic surface texture, softness, and surface liquid properties
so that the pre-trained CNN architecture can be transferred confidently into a
real endoscopic scenario. An average error of 7:1% and 3:4% for translation and
rotation has been obtained, respectively. The results accomplished from the
experiments demonstrate that a CNN pre-trained with raw 2D endoscopic images
performs accurately inside the GI tract and is robust to various challenges
posed by reflection distortions, lens imperfections, vignetting, noise, motion
blur, low resolution, and lack of unique landmarks to track.
| new_dataset | 0.87584 |
1705.05455 | Saad Bin Ahmed | Saad Bin Ahmed, Saeeda Naz, Salahuddin Swati, Muhammad Imran Razzak | Handwritten Urdu Character Recognition using 1-Dimensional BLSTM
Classifier | 10 pages, Accepted in NCA for publication | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recognition of cursive script is regarded as a subtle task in optical
character recognition due to its varied representation. Every cursive script
has different nature and associated challenges. As Urdu is one of cursive
language that is derived from Arabic script, thats why it nearly shares the
same challenges and difficulties even more harder. We can categorized Urdu and
Arabic language on basis of its script they use. Urdu is mostly written in
Nastaliq style whereas, Arabic follows Naskh style of writing. This paper
presents new and comprehensive Urdu handwritten offline database name
Urdu-Nastaliq Handwritten Dataset (UNHD). Currently, there is no standard and
comprehensive Urdu handwritten dataset available publicly for researchers. The
acquired dataset covers commonly used ligatures that were written by 500
writers with their natural handwriting on A4 size paper. We performed
experiments using recurrent neural networks and reported a significant accuracy
for handwritten Urdu character recognition.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 21:13:08 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Ahmed",
"Saad Bin",
""
],
[
"Naz",
"Saeeda",
""
],
[
"Swati",
"Salahuddin",
""
],
[
"Razzak",
"Muhammad Imran",
""
]
] | TITLE: Handwritten Urdu Character Recognition using 1-Dimensional BLSTM
Classifier
ABSTRACT: The recognition of cursive script is regarded as a subtle task in optical
character recognition due to its varied representation. Every cursive script
has different nature and associated challenges. As Urdu is one of cursive
language that is derived from Arabic script, thats why it nearly shares the
same challenges and difficulties even more harder. We can categorized Urdu and
Arabic language on basis of its script they use. Urdu is mostly written in
Nastaliq style whereas, Arabic follows Naskh style of writing. This paper
presents new and comprehensive Urdu handwritten offline database name
Urdu-Nastaliq Handwritten Dataset (UNHD). Currently, there is no standard and
comprehensive Urdu handwritten dataset available publicly for researchers. The
acquired dataset covers commonly used ligatures that were written by 500
writers with their natural handwriting on A4 size paper. We performed
experiments using recurrent neural networks and reported a significant accuracy
for handwritten Urdu character recognition.
| new_dataset | 0.957636 |
1705.05483 | Andrei Polzounov | Andrei Polzounov, Artsiom Ablavatski, Sergio Escalera, Shijian Lu,
Jianfei Cai | WordFence: Text Detection in Natural Images with Border Awareness | 5 pages, 2 figures, ICIP 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, text recognition has achieved remarkable success in
recognizing scanned document text. However, word recognition in natural images
is still an open problem, which generally requires time consuming
post-processing steps. We present a novel architecture for individual word
detection in scene images based on semantic segmentation. Our contributions are
twofold: the concept of WordFence, which detects border areas surrounding each
individual word and a novel pixelwise weighted softmax loss function which
penalizes background and emphasizes small text regions. WordFence ensures that
each word is detected individually, and the new loss function provides a strong
training signal to both text and word border localization. The proposed
technique avoids intensive post-processing, producing an end-to-end word
detection system. We achieve superior localization recall on common benchmark
datasets - 92% recall on ICDAR11 and ICDAR13 and 63% recall on SVT.
Furthermore, our end-to-end word recognition system achieves state-of-the-art
86% F-Score on ICDAR13.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 23:42:59 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Polzounov",
"Andrei",
""
],
[
"Ablavatski",
"Artsiom",
""
],
[
"Escalera",
"Sergio",
""
],
[
"Lu",
"Shijian",
""
],
[
"Cai",
"Jianfei",
""
]
] | TITLE: WordFence: Text Detection in Natural Images with Border Awareness
ABSTRACT: In recent years, text recognition has achieved remarkable success in
recognizing scanned document text. However, word recognition in natural images
is still an open problem, which generally requires time consuming
post-processing steps. We present a novel architecture for individual word
detection in scene images based on semantic segmentation. Our contributions are
twofold: the concept of WordFence, which detects border areas surrounding each
individual word and a novel pixelwise weighted softmax loss function which
penalizes background and emphasizes small text regions. WordFence ensures that
each word is detected individually, and the new loss function provides a strong
training signal to both text and word border localization. The proposed
technique avoids intensive post-processing, producing an end-to-end word
detection system. We achieve superior localization recall on common benchmark
datasets - 92% recall on ICDAR11 and ICDAR13 and 63% recall on SVT.
Furthermore, our end-to-end word recognition system achieves state-of-the-art
86% F-Score on ICDAR13.
| no_new_dataset | 0.953966 |
1705.05494 | Paulo Roberto Urio | Paulo Roberto Urio, Zhao Liang | Data clustering with edge domination in complex networks | 13 pages, 6 figures | null | null | null | cs.SI cs.LG physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a model for a dynamical system where particles dominate
edges in a complex network. The proposed dynamical system is then extended to
an application on the problem of community detection and data clustering. In
the case of the data clustering problem, 6 different techniques were simulated
on 10 different datasets in order to compare with the proposed technique. The
results show that the proposed algorithm performs well when prior knowledge of
the number of clusters is known to the algorithm.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 00:44:31 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Urio",
"Paulo Roberto",
""
],
[
"Liang",
"Zhao",
""
]
] | TITLE: Data clustering with edge domination in complex networks
ABSTRACT: This paper presents a model for a dynamical system where particles dominate
edges in a complex network. The proposed dynamical system is then extended to
an application on the problem of community detection and data clustering. In
the case of the data clustering problem, 6 different techniques were simulated
on 10 different datasets in order to compare with the proposed technique. The
results show that the proposed algorithm performs well when prior knowledge of
the number of clusters is known to the algorithm.
| no_new_dataset | 0.949342 |
1705.05498 | Jing Zhang | Jing Zhang and Wanqing Li and Philip Ogunbona | Joint Geometrical and Statistical Alignment for Visual Domain Adaptation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel unsupervised domain adaptation method for
cross-domain visual recognition. We propose a unified framework that reduces
the shift between domains both statistically and geometrically, referred to as
Joint Geometrical and Statistical Alignment (JGSA). Specifically, we learn two
coupled projections that project the source domain and target domain data into
low dimensional subspaces where the geometrical shift and distribution shift
are reduced simultaneously. The objective function can be solved efficiently in
a closed form. Extensive experiments have verified that the proposed method
significantly outperforms several state-of-the-art domain adaptation methods on
a synthetic dataset and three different real world cross-domain visual
recognition tasks.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 01:35:58 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Zhang",
"Jing",
""
],
[
"Li",
"Wanqing",
""
],
[
"Ogunbona",
"Philip",
""
]
] | TITLE: Joint Geometrical and Statistical Alignment for Visual Domain Adaptation
ABSTRACT: This paper presents a novel unsupervised domain adaptation method for
cross-domain visual recognition. We propose a unified framework that reduces
the shift between domains both statistically and geometrically, referred to as
Joint Geometrical and Statistical Alignment (JGSA). Specifically, we learn two
coupled projections that project the source domain and target domain data into
low dimensional subspaces where the geometrical shift and distribution shift
are reduced simultaneously. The objective function can be solved efficiently in
a closed form. Extensive experiments have verified that the proposed method
significantly outperforms several state-of-the-art domain adaptation methods on
a synthetic dataset and three different real world cross-domain visual
recognition tasks.
| no_new_dataset | 0.951594 |
1705.05508 | Yong Khoo | Yong Khoo, Sang Chung | Automated Body Structure Extraction from Arbitrary 3D Mesh | null | Imaging and Graphics, 2017 | null | null | cs.GR cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents an automated method for 3D character skeleton extraction
that can be applied for generic 3D shapes. Our work is motivated by the
skeleton-based prior work on automatic rigging focused on skeleton extraction
and can automatically aligns the extracted structure to fit the 3D shape of the
given 3D mesh. The body mesh can be subsequently skinned based on the extracted
skeleton and thus enables rigging process. In the experiment, we apply public
dataset to drive the estimated skeleton from different body shapes, as well as
the real data obtained from 3D scanning systems. Satisfactory results are
obtained compared to the existing approaches.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 02:58:44 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Khoo",
"Yong",
""
],
[
"Chung",
"Sang",
""
]
] | TITLE: Automated Body Structure Extraction from Arbitrary 3D Mesh
ABSTRACT: This paper presents an automated method for 3D character skeleton extraction
that can be applied for generic 3D shapes. Our work is motivated by the
skeleton-based prior work on automatic rigging focused on skeleton extraction
and can automatically aligns the extracted structure to fit the 3D shape of the
given 3D mesh. The body mesh can be subsequently skinned based on the extracted
skeleton and thus enables rigging process. In the experiment, we apply public
dataset to drive the estimated skeleton from different body shapes, as well as
the real data obtained from 3D scanning systems. Satisfactory results are
obtained compared to the existing approaches.
| no_new_dataset | 0.954308 |
1705.05592 | Varun Ojha | Varun Kumar Ojha, Ajith Abraham, V\'aclav Sn\'a\v{s}el | Ensemble of heterogeneous flexible neural trees using multiobjective
genetic programming | null | Applied Soft Computing, 2017, Volume 52 Pages 909 to 924 | 10.1016/j.asoc.2016.09.035 | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Machine learning algorithms are inherently multiobjective in nature, where
approximation error minimization and model's complexity simplification are two
conflicting objectives. We proposed a multiobjective genetic programming (MOGP)
for creating a heterogeneous flexible neural tree (HFNT), tree-like flexible
feedforward neural network model. The functional heterogeneity in neural tree
nodes was introduced to capture a better insight of data during learning
because each input in a dataset possess different features. MOGP guided an
initial HFNT population towards Pareto-optimal solutions, where the final
population was used for making an ensemble system. A diversity index measure
along with approximation error and complexity was introduced to maintain
diversity among the candidates in the population. Hence, the ensemble was
created by using accurate, structurally simple, and diverse candidates from
MOGP final population. Differential evolution algorithm was applied to
fine-tune the underlying parameters of the selected candidates. A comprehensive
test over classification, regression, and time-series datasets proved the
efficiency of the proposed algorithm over other available prediction methods.
Moreover, the heterogeneous creation of HFNT proved to be efficient in making
ensemble system from the final population.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 08:40:42 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Ojha",
"Varun Kumar",
""
],
[
"Abraham",
"Ajith",
""
],
[
"Snášel",
"Václav",
""
]
] | TITLE: Ensemble of heterogeneous flexible neural trees using multiobjective
genetic programming
ABSTRACT: Machine learning algorithms are inherently multiobjective in nature, where
approximation error minimization and model's complexity simplification are two
conflicting objectives. We proposed a multiobjective genetic programming (MOGP)
for creating a heterogeneous flexible neural tree (HFNT), tree-like flexible
feedforward neural network model. The functional heterogeneity in neural tree
nodes was introduced to capture a better insight of data during learning
because each input in a dataset possess different features. MOGP guided an
initial HFNT population towards Pareto-optimal solutions, where the final
population was used for making an ensemble system. A diversity index measure
along with approximation error and complexity was introduced to maintain
diversity among the candidates in the population. Hence, the ensemble was
created by using accurate, structurally simple, and diverse candidates from
MOGP final population. Differential evolution algorithm was applied to
fine-tune the underlying parameters of the selected candidates. A comprehensive
test over classification, regression, and time-series datasets proved the
efficiency of the proposed algorithm over other available prediction methods.
Moreover, the heterogeneous creation of HFNT proved to be efficient in making
ensemble system from the final population.
| no_new_dataset | 0.953319 |
1705.05640 | Limin Wang | Wen Li, Limin Wang, Wei Li, Eirikur Agustsson, Jesse Berent, Abhinav
Gupta, Rahul Sukthankar, Luc Van Gool | WebVision Challenge: Visual Learning and Understanding With Web Data | project page: http://www.vision.ee.ethz.ch/webvision/ | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present the 2017 WebVision Challenge, a public image recognition challenge
designed for deep learning based on web images without instance-level human
annotation. Following the spirit of previous vision challenges, such as ILSVRC,
Places2 and PASCAL VOC, which have played critical roles in the development of
computer vision by contributing to the community with large scale annotated
data for model designing and standardized benchmarking, we contribute with this
challenge a large scale web images dataset, and a public competition with a
workshop co-located with CVPR 2017. The WebVision dataset contains more than
$2.4$ million web images crawled from the Internet by using queries generated
from the $1,000$ semantic concepts of the benchmark ILSVRC 2012 dataset. Meta
information is also included. A validation set and test set containing human
annotated images are also provided to facilitate algorithmic development. The
2017 WebVision challenge consists of two tracks, the image classification task
on WebVision test set, and the transfer learning task on PASCAL VOC 2012
dataset. In this paper, we describe the details of data collection and
annotation, highlight the characteristics of the dataset, and introduce the
evaluation metrics.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 10:59:23 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Li",
"Wen",
""
],
[
"Wang",
"Limin",
""
],
[
"Li",
"Wei",
""
],
[
"Agustsson",
"Eirikur",
""
],
[
"Berent",
"Jesse",
""
],
[
"Gupta",
"Abhinav",
""
],
[
"Sukthankar",
"Rahul",
""
],
[
"Van Gool",
"Luc",
""
]
] | TITLE: WebVision Challenge: Visual Learning and Understanding With Web Data
ABSTRACT: We present the 2017 WebVision Challenge, a public image recognition challenge
designed for deep learning based on web images without instance-level human
annotation. Following the spirit of previous vision challenges, such as ILSVRC,
Places2 and PASCAL VOC, which have played critical roles in the development of
computer vision by contributing to the community with large scale annotated
data for model designing and standardized benchmarking, we contribute with this
challenge a large scale web images dataset, and a public competition with a
workshop co-located with CVPR 2017. The WebVision dataset contains more than
$2.4$ million web images crawled from the Internet by using queries generated
from the $1,000$ semantic concepts of the benchmark ILSVRC 2012 dataset. Meta
information is also included. A validation set and test set containing human
annotated images are also provided to facilitate algorithmic development. The
2017 WebVision challenge consists of two tracks, the image classification task
on WebVision test set, and the transfer learning task on PASCAL VOC 2012
dataset. In this paper, we describe the details of data collection and
annotation, highlight the characteristics of the dataset, and introduce the
evaluation metrics.
| new_dataset | 0.938181 |
1705.05756 | Witold Rudnicki | Krzysztof Mnich and Witold R. Rudnicki | All-relevant feature selection using multidimensional filters with
exhaustive search | 27 pages, 11 figures, 3 tables | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper describes a method for identification of the informative variables
in the information system with discrete decision variables. It is targeted
specifically towards discovery of the variables that are non-informative when
considered alone, but are informative when the synergistic interactions between
multiple variables are considered. To this end, the mutual entropy of all
possible k-tuples of variables with decision variable is computed. Then, for
each variable the maximal information gain due to interactions with other
variables is obtained. For non-informative variables this quantity conforms to
the well known statistical distributions. This allows for discerning truly
informative variables from non-informative ones. For demonstration of the
approach, the method is applied to several synthetic datasets that involve
complex multidimensional interactions between variables. It is capable of
identifying most important informative variables, even in the case when the
dimensionality of the analysis is smaller than the true dimensionality of the
problem. What is more, the high sensitivity of the algorithm allows for
detection of the influence of nuisance variables on the response variable.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 15:11:10 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Mnich",
"Krzysztof",
""
],
[
"Rudnicki",
"Witold R.",
""
]
] | TITLE: All-relevant feature selection using multidimensional filters with
exhaustive search
ABSTRACT: This paper describes a method for identification of the informative variables
in the information system with discrete decision variables. It is targeted
specifically towards discovery of the variables that are non-informative when
considered alone, but are informative when the synergistic interactions between
multiple variables are considered. To this end, the mutual entropy of all
possible k-tuples of variables with decision variable is computed. Then, for
each variable the maximal information gain due to interactions with other
variables is obtained. For non-informative variables this quantity conforms to
the well known statistical distributions. This allows for discerning truly
informative variables from non-informative ones. For demonstration of the
approach, the method is applied to several synthetic datasets that involve
complex multidimensional interactions between variables. It is capable of
identifying most important informative variables, even in the case when the
dimensionality of the analysis is smaller than the true dimensionality of the
problem. What is more, the high sensitivity of the algorithm allows for
detection of the influence of nuisance variables on the response variable.
| no_new_dataset | 0.9463 |
1705.05787 | Luiz Gustavo Hafemann | Luiz G. Hafemann, Robert Sabourin, Luiz S. Oliveira | Learning Features for Offline Handwritten Signature Verification using
Deep Convolutional Neural Networks | null | null | 10.1016/j.patcog.2017.05.012 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Verifying the identity of a person using handwritten signatures is
challenging in the presence of skilled forgeries, where a forger has access to
a person's signature and deliberately attempt to imitate it. In offline
(static) signature verification, the dynamic information of the signature
writing process is lost, and it is difficult to design good feature extractors
that can distinguish genuine signatures and skilled forgeries. This reflects in
a relatively poor performance, with verification errors around 7% in the best
systems in the literature. To address both the difficulty of obtaining good
features, as well as improve system performance, we propose learning the
representations from signature images, in a Writer-Independent format, using
Convolutional Neural Networks. In particular, we propose a novel formulation of
the problem that includes knowledge of skilled forgeries from a subset of users
in the feature learning process, that aims to capture visual cues that
distinguish genuine signatures and forgeries regardless of the user. Extensive
experiments were conducted on four datasets: GPDS, MCYT, CEDAR and Brazilian
PUC-PR datasets. On GPDS-160, we obtained a large improvement in
state-of-the-art performance, achieving 1.72% Equal Error Rate, compared to
6.97% in the literature. We also verified that the features generalize beyond
the GPDS dataset, surpassing the state-of-the-art performance in the other
datasets, without requiring the representation to be fine-tuned to each
particular dataset.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 16:08:09 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Hafemann",
"Luiz G.",
""
],
[
"Sabourin",
"Robert",
""
],
[
"Oliveira",
"Luiz S.",
""
]
] | TITLE: Learning Features for Offline Handwritten Signature Verification using
Deep Convolutional Neural Networks
ABSTRACT: Verifying the identity of a person using handwritten signatures is
challenging in the presence of skilled forgeries, where a forger has access to
a person's signature and deliberately attempt to imitate it. In offline
(static) signature verification, the dynamic information of the signature
writing process is lost, and it is difficult to design good feature extractors
that can distinguish genuine signatures and skilled forgeries. This reflects in
a relatively poor performance, with verification errors around 7% in the best
systems in the literature. To address both the difficulty of obtaining good
features, as well as improve system performance, we propose learning the
representations from signature images, in a Writer-Independent format, using
Convolutional Neural Networks. In particular, we propose a novel formulation of
the problem that includes knowledge of skilled forgeries from a subset of users
in the feature learning process, that aims to capture visual cues that
distinguish genuine signatures and forgeries regardless of the user. Extensive
experiments were conducted on four datasets: GPDS, MCYT, CEDAR and Brazilian
PUC-PR datasets. On GPDS-160, we obtained a large improvement in
state-of-the-art performance, achieving 1.72% Equal Error Rate, compared to
6.97% in the literature. We also verified that the features generalize beyond
the GPDS dataset, surpassing the state-of-the-art performance in the other
datasets, without requiring the representation to be fine-tuned to each
particular dataset.
| no_new_dataset | 0.946892 |
1705.05823 | Oren Rippel | Oren Rippel, Lubomir Bourdev | Real-Time Adaptive Image Compression | Published at ICML 2017 | null | null | null | stat.ML cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a machine learning-based approach to lossy image compression which
outperforms all existing codecs, while running in real-time.
Our algorithm typically produces files 2.5 times smaller than JPEG and JPEG
2000, 2 times smaller than WebP, and 1.7 times smaller than BPG on datasets of
generic images across all quality levels. At the same time, our codec is
designed to be lightweight and deployable: for example, it can encode or decode
the Kodak dataset in around 10ms per image on GPU.
Our architecture is an autoencoder featuring pyramidal analysis, an adaptive
coding module, and regularization of the expected codelength. We also
supplement our approach with adversarial training specialized towards use in a
compression setting: this enables us to produce visually pleasing
reconstructions for very low bitrates.
| [
{
"version": "v1",
"created": "Tue, 16 May 2017 17:51:07 GMT"
}
] | 2017-05-17T00:00:00 | [
[
"Rippel",
"Oren",
""
],
[
"Bourdev",
"Lubomir",
""
]
] | TITLE: Real-Time Adaptive Image Compression
ABSTRACT: We present a machine learning-based approach to lossy image compression which
outperforms all existing codecs, while running in real-time.
Our algorithm typically produces files 2.5 times smaller than JPEG and JPEG
2000, 2 times smaller than WebP, and 1.7 times smaller than BPG on datasets of
generic images across all quality levels. At the same time, our codec is
designed to be lightweight and deployable: for example, it can encode or decode
the Kodak dataset in around 10ms per image on GPU.
Our architecture is an autoencoder featuring pyramidal analysis, an adaptive
coding module, and regularization of the expected codelength. We also
supplement our approach with adversarial training specialized towards use in a
compression setting: this enables us to produce visually pleasing
reconstructions for very low bitrates.
| no_new_dataset | 0.940408 |
1512.07797 | Jiaqian Yu | Jiaqian Yu (CVC, GALEN), Matthew Blaschko | The Lov\'asz Hinge: A Novel Convex Surrogate for Submodular Losses | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning with non-modular losses is an important problem when sets of
predictions are made simultaneously. The main tools for constructing convex
surrogate loss functions for set prediction are margin rescaling and slack
rescaling. In this work, we show that these strategies lead to tight convex
surrogates iff the underlying loss function is increasing in the number of
incorrect predictions. However, gradient or cutting-plane computation for these
functions is NP-hard for non-supermodular loss functions. We propose instead a
novel surrogate loss function for submodular losses, the Lov\'asz hinge, which
leads to O(p log p) complexity with O(p) oracle accesses to the loss function
to compute a gradient or cutting-plane. We prove that the Lov\'asz hinge is
convex and yields an extension. As a result, we have developed the first
tractable convex surrogates in the literature for submodular losses. We
demonstrate the utility of this novel convex surrogate through several set
prediction tasks, including on the PASCAL VOC and Microsoft COCO datasets.
| [
{
"version": "v1",
"created": "Thu, 24 Dec 2015 11:49:47 GMT"
},
{
"version": "v2",
"created": "Mon, 15 May 2017 11:25:31 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Yu",
"Jiaqian",
"",
"CVC, GALEN"
],
[
"Blaschko",
"Matthew",
""
]
] | TITLE: The Lov\'asz Hinge: A Novel Convex Surrogate for Submodular Losses
ABSTRACT: Learning with non-modular losses is an important problem when sets of
predictions are made simultaneously. The main tools for constructing convex
surrogate loss functions for set prediction are margin rescaling and slack
rescaling. In this work, we show that these strategies lead to tight convex
surrogates iff the underlying loss function is increasing in the number of
incorrect predictions. However, gradient or cutting-plane computation for these
functions is NP-hard for non-supermodular loss functions. We propose instead a
novel surrogate loss function for submodular losses, the Lov\'asz hinge, which
leads to O(p log p) complexity with O(p) oracle accesses to the loss function
to compute a gradient or cutting-plane. We prove that the Lov\'asz hinge is
convex and yields an extension. As a result, we have developed the first
tractable convex surrogates in the literature for submodular losses. We
demonstrate the utility of this novel convex surrogate through several set
prediction tasks, including on the PASCAL VOC and Microsoft COCO datasets.
| no_new_dataset | 0.951997 |
1608.04267 | Zihan Zhou | Zihan Zhou, Farshid Farhat, James Z. Wang | Detecting Dominant Vanishing Points in Natural Scenes with Application
to Composition-Sensitive Image Retrieval | 15 pages, 18 figures, to appear in IEEE Transactions on Multimedia | null | null | null | cs.CV cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Linear perspective is widely used in landscape photography to create the
impression of depth on a 2D photo. Automated understanding of linear
perspective in landscape photography has several real-world applications,
including aesthetics assessment, image retrieval, and on-site feedback for
photo composition, yet adequate automated understanding has been elusive. We
address this problem by detecting the dominant vanishing point and the
associated line structures in a photo. However, natural landscape scenes pose
great technical challenges because often the inadequate number of strong edges
converging to the dominant vanishing point is inadequate. To overcome this
difficulty, we propose a novel vanishing point detection method that exploits
global structures in the scene via contour detection. We show that our method
significantly outperforms state-of-the-art methods on a public ground truth
landscape image dataset that we have created. Based on the detection results,
we further demonstrate how our approach to linear perspective understanding
provides on-site guidance to amateur photographers on their work through a
novel viewpoint-specific image retrieval system.
| [
{
"version": "v1",
"created": "Mon, 15 Aug 2016 13:48:22 GMT"
},
{
"version": "v2",
"created": "Sat, 13 May 2017 14:58:05 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Zhou",
"Zihan",
""
],
[
"Farhat",
"Farshid",
""
],
[
"Wang",
"James Z.",
""
]
] | TITLE: Detecting Dominant Vanishing Points in Natural Scenes with Application
to Composition-Sensitive Image Retrieval
ABSTRACT: Linear perspective is widely used in landscape photography to create the
impression of depth on a 2D photo. Automated understanding of linear
perspective in landscape photography has several real-world applications,
including aesthetics assessment, image retrieval, and on-site feedback for
photo composition, yet adequate automated understanding has been elusive. We
address this problem by detecting the dominant vanishing point and the
associated line structures in a photo. However, natural landscape scenes pose
great technical challenges because often the inadequate number of strong edges
converging to the dominant vanishing point is inadequate. To overcome this
difficulty, we propose a novel vanishing point detection method that exploits
global structures in the scene via contour detection. We show that our method
significantly outperforms state-of-the-art methods on a public ground truth
landscape image dataset that we have created. Based on the detection results,
we further demonstrate how our approach to linear perspective understanding
provides on-site guidance to amateur photographers on their work through a
novel viewpoint-specific image retrieval system.
| new_dataset | 0.959762 |
1612.00837 | Yash Goyal | Yash Goyal, Tejas Khot, Douglas Summers-Stay, Dhruv Batra, Devi Parikh | Making the V in VQA Matter: Elevating the Role of Image Understanding in
Visual Question Answering | null | null | null | null | cs.CV cs.AI cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Problems at the intersection of vision and language are of significant
importance both as challenging research questions and for the rich set of
applications they enable. However, inherent structure in our world and bias in
our language tend to be a simpler signal for learning than visual modalities,
resulting in models that ignore visual information, leading to an inflated
sense of their capability.
We propose to counter these language priors for the task of Visual Question
Answering (VQA) and make vision (the V in VQA) matter! Specifically, we balance
the popular VQA dataset by collecting complementary images such that every
question in our balanced dataset is associated with not just a single image,
but rather a pair of similar images that result in two different answers to the
question. Our dataset is by construction more balanced than the original VQA
dataset and has approximately twice the number of image-question pairs. Our
complete balanced dataset is publicly available at www.visualqa.org as part of
the 2nd iteration of the Visual Question Answering Dataset and Challenge (VQA
v2.0).
We further benchmark a number of state-of-art VQA models on our balanced
dataset. All models perform significantly worse on our balanced dataset,
suggesting that these models have indeed learned to exploit language priors.
This finding provides the first concrete empirical evidence for what seems to
be a qualitative sense among practitioners.
Finally, our data collection protocol for identifying complementary images
enables us to develop a novel interpretable model, which in addition to
providing an answer to the given (image, question) pair, also provides a
counter-example based explanation. Specifically, it identifies an image that is
similar to the original image, but it believes has a different answer to the
same question. This can help in building trust for machines among their users.
| [
{
"version": "v1",
"created": "Fri, 2 Dec 2016 20:57:07 GMT"
},
{
"version": "v2",
"created": "Fri, 14 Apr 2017 18:20:13 GMT"
},
{
"version": "v3",
"created": "Mon, 15 May 2017 17:58:49 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Goyal",
"Yash",
""
],
[
"Khot",
"Tejas",
""
],
[
"Summers-Stay",
"Douglas",
""
],
[
"Batra",
"Dhruv",
""
],
[
"Parikh",
"Devi",
""
]
] | TITLE: Making the V in VQA Matter: Elevating the Role of Image Understanding in
Visual Question Answering
ABSTRACT: Problems at the intersection of vision and language are of significant
importance both as challenging research questions and for the rich set of
applications they enable. However, inherent structure in our world and bias in
our language tend to be a simpler signal for learning than visual modalities,
resulting in models that ignore visual information, leading to an inflated
sense of their capability.
We propose to counter these language priors for the task of Visual Question
Answering (VQA) and make vision (the V in VQA) matter! Specifically, we balance
the popular VQA dataset by collecting complementary images such that every
question in our balanced dataset is associated with not just a single image,
but rather a pair of similar images that result in two different answers to the
question. Our dataset is by construction more balanced than the original VQA
dataset and has approximately twice the number of image-question pairs. Our
complete balanced dataset is publicly available at www.visualqa.org as part of
the 2nd iteration of the Visual Question Answering Dataset and Challenge (VQA
v2.0).
We further benchmark a number of state-of-art VQA models on our balanced
dataset. All models perform significantly worse on our balanced dataset,
suggesting that these models have indeed learned to exploit language priors.
This finding provides the first concrete empirical evidence for what seems to
be a qualitative sense among practitioners.
Finally, our data collection protocol for identifying complementary images
enables us to develop a novel interpretable model, which in addition to
providing an answer to the given (image, question) pair, also provides a
counter-example based explanation. Specifically, it identifies an image that is
similar to the original image, but it believes has a different answer to the
same question. This can help in building trust for machines among their users.
| new_dataset | 0.921781 |
1612.01414 | Alexander Jung | Alexander Jung, Alfred O. Hero III, Alexandru Mara, and Saeed Jahromi | Semi-Supervised Learning via Sparse Label Propagation | null | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work proposes a novel method for semi-supervised learning from partially
labeled massive network-structured datasets, i.e., big data over networks. We
model the underlying hypothesis, which relates data points to labels, as a
graph signal, defined over some graph (network) structure intrinsic to the
dataset. Following the key principle of supervised learning, i.e., similar
inputs yield similar outputs, we require the graph signals induced by labels to
have small total variation. Accordingly, we formulate the problem of learning
the labels of data points as a non-smooth convex optimization problem which
amounts to balancing between the empirical loss, i.e., the discrepancy with
some partially available label information, and the smoothness quantified by
the total variation of the learned graph signal. We solve this optimization
problem by appealing to a recently proposed preconditioned variant of the
popular primal-dual method by Pock and Chambolle, which results in a sparse
label propagation algorithm. This learning algorithm allows for a highly
scalable implementation as message passing over the underlying data graph. By
applying concepts of compressed sensing to the learning problem, we are also
able to provide a transparent sufficient condition on the underlying network
structure such that accurate learning of the labels is possible. We also
present an implementation of the message passing formulation allows for a
highly scalable implementation in big data frameworks.
| [
{
"version": "v1",
"created": "Mon, 5 Dec 2016 16:04:38 GMT"
},
{
"version": "v2",
"created": "Thu, 22 Dec 2016 15:41:31 GMT"
},
{
"version": "v3",
"created": "Wed, 10 May 2017 16:57:05 GMT"
},
{
"version": "v4",
"created": "Mon, 15 May 2017 07:53:13 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Jung",
"Alexander",
""
],
[
"Hero",
"Alfred O.",
"III"
],
[
"Mara",
"Alexandru",
""
],
[
"Jahromi",
"Saeed",
""
]
] | TITLE: Semi-Supervised Learning via Sparse Label Propagation
ABSTRACT: This work proposes a novel method for semi-supervised learning from partially
labeled massive network-structured datasets, i.e., big data over networks. We
model the underlying hypothesis, which relates data points to labels, as a
graph signal, defined over some graph (network) structure intrinsic to the
dataset. Following the key principle of supervised learning, i.e., similar
inputs yield similar outputs, we require the graph signals induced by labels to
have small total variation. Accordingly, we formulate the problem of learning
the labels of data points as a non-smooth convex optimization problem which
amounts to balancing between the empirical loss, i.e., the discrepancy with
some partially available label information, and the smoothness quantified by
the total variation of the learned graph signal. We solve this optimization
problem by appealing to a recently proposed preconditioned variant of the
popular primal-dual method by Pock and Chambolle, which results in a sparse
label propagation algorithm. This learning algorithm allows for a highly
scalable implementation as message passing over the underlying data graph. By
applying concepts of compressed sensing to the learning problem, we are also
able to provide a transparent sufficient condition on the underlying network
structure such that accurate learning of the labels is possible. We also
present an implementation of the message passing formulation allows for a
highly scalable implementation in big data frameworks.
| no_new_dataset | 0.948298 |
1702.08400 | Kuniaki Saito Saito Kuniaki | Kuniaki Saito, Yoshitaka Ushiku and Tatsuya Harada | Asymmetric Tri-training for Unsupervised Domain Adaptation | TBA on ICML2017 | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep-layered models trained on a large number of labeled samples boost the
accuracy of many tasks. It is important to apply such models to different
domains because collecting many labeled samples in various domains is
expensive. In unsupervised domain adaptation, one needs to train a classifier
that works well on a target domain when provided with labeled source samples
and unlabeled target samples. Although many methods aim to match the
distributions of source and target samples, simply matching the distribution
cannot ensure accuracy on the target domain. To learn discriminative
representations for the target domain, we assume that artificially labeling
target samples can result in a good representation. Tri-training leverages
three classifiers equally to give pseudo-labels to unlabeled samples, but the
method does not assume labeling samples generated from a different domain.In
this paper, we propose an asymmetric tri-training method for unsupervised
domain adaptation, where we assign pseudo-labels to unlabeled samples and train
neural networks as if they are true labels. In our work, we use three networks
asymmetrically. By asymmetric, we mean that two networks are used to label
unlabeled target samples and one network is trained by the samples to obtain
target-discriminative representations. We evaluate our method on digit
recognition and sentiment analysis datasets. Our proposed method achieves
state-of-the-art performance on the benchmark digit recognition datasets of
domain adaptation.
| [
{
"version": "v1",
"created": "Mon, 27 Feb 2017 17:48:17 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 15:11:14 GMT"
},
{
"version": "v3",
"created": "Sat, 13 May 2017 05:44:03 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Saito",
"Kuniaki",
""
],
[
"Ushiku",
"Yoshitaka",
""
],
[
"Harada",
"Tatsuya",
""
]
] | TITLE: Asymmetric Tri-training for Unsupervised Domain Adaptation
ABSTRACT: Deep-layered models trained on a large number of labeled samples boost the
accuracy of many tasks. It is important to apply such models to different
domains because collecting many labeled samples in various domains is
expensive. In unsupervised domain adaptation, one needs to train a classifier
that works well on a target domain when provided with labeled source samples
and unlabeled target samples. Although many methods aim to match the
distributions of source and target samples, simply matching the distribution
cannot ensure accuracy on the target domain. To learn discriminative
representations for the target domain, we assume that artificially labeling
target samples can result in a good representation. Tri-training leverages
three classifiers equally to give pseudo-labels to unlabeled samples, but the
method does not assume labeling samples generated from a different domain.In
this paper, we propose an asymmetric tri-training method for unsupervised
domain adaptation, where we assign pseudo-labels to unlabeled samples and train
neural networks as if they are true labels. In our work, we use three networks
asymmetrically. By asymmetric, we mean that two networks are used to label
unlabeled target samples and one network is trained by the samples to obtain
target-discriminative representations. We evaluate our method on digit
recognition and sentiment analysis datasets. Our proposed method achieves
state-of-the-art performance on the benchmark digit recognition datasets of
domain adaptation.
| no_new_dataset | 0.950411 |
1703.01289 | Sebastian Bullinger | Sebastian Bullinger, Christoph Bodensteiner and Michael Arens | Instance Flow Based Online Multiple Object Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a method to perform online Multiple Object Tracking (MOT) of known
object categories in monocular video data. Current Tracking-by-Detection MOT
approaches build on top of 2D bounding box detections. In contrast, we exploit
state-of-the-art instance aware semantic segmentation techniques to compute 2D
shape representations of target objects in each frame. We predict position and
shape of segmented instances in subsequent frames by exploiting optical flow
cues. We define an affinity matrix between instances of subsequent frames which
reflects locality and visual similarity. The instance association is solved by
applying the Hungarian method. We evaluate different configurations of our
algorithm using the MOT 2D 2015 train dataset. The evaluation shows that our
tracking approach is able to track objects with high relative motions. In
addition, we provide results of our approach on the MOT 2D 2015 test set for
comparison with previous works. We achieve a MOTA score of 32.1.
| [
{
"version": "v1",
"created": "Fri, 3 Mar 2017 18:54:55 GMT"
},
{
"version": "v2",
"created": "Mon, 15 May 2017 14:14:30 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Bullinger",
"Sebastian",
""
],
[
"Bodensteiner",
"Christoph",
""
],
[
"Arens",
"Michael",
""
]
] | TITLE: Instance Flow Based Online Multiple Object Tracking
ABSTRACT: We present a method to perform online Multiple Object Tracking (MOT) of known
object categories in monocular video data. Current Tracking-by-Detection MOT
approaches build on top of 2D bounding box detections. In contrast, we exploit
state-of-the-art instance aware semantic segmentation techniques to compute 2D
shape representations of target objects in each frame. We predict position and
shape of segmented instances in subsequent frames by exploiting optical flow
cues. We define an affinity matrix between instances of subsequent frames which
reflects locality and visual similarity. The instance association is solved by
applying the Hungarian method. We evaluate different configurations of our
algorithm using the MOT 2D 2015 train dataset. The evaluation shows that our
tracking approach is able to track objects with high relative motions. In
addition, we provide results of our approach on the MOT 2D 2015 test set for
comparison with previous works. We achieve a MOTA score of 32.1.
| no_new_dataset | 0.948058 |
1705.00274 | Sibel Tari | Asli Genctav, Yusuf Sahillioglu, and Sibel Tari | Topologically Robust 3D Shape Matching via Gradual Deflation and
Inflation | Section 2 replaced | null | null | null | cs.GR cs.CG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite being vastly ignored in the literature, coping with topological noise
is an issue of increasing importance, especially as a consequence of the
increasing number and diversity of 3D polygonal models that are captured by
devices of different qualities or synthesized by algorithms of different
stabilities. One approach for matching 3D shapes under topological noise is to
replace the topology-sensitive geodesic distance with distances that are less
sensitive to topological changes. We propose an alternative approach utilising
gradual deflation (or inflation) of the shape volume, of which purpose is to
bring the pair of shapes to be matched to a \emph{comparable} topology before
the search for correspondences. Illustrative experiments using different
datasets demonstrate that as the level of topological noise increases, our
approach outperforms the other methods in the literature.
| [
{
"version": "v1",
"created": "Sun, 30 Apr 2017 06:40:18 GMT"
},
{
"version": "v2",
"created": "Fri, 12 May 2017 21:48:29 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Genctav",
"Asli",
""
],
[
"Sahillioglu",
"Yusuf",
""
],
[
"Tari",
"Sibel",
""
]
] | TITLE: Topologically Robust 3D Shape Matching via Gradual Deflation and
Inflation
ABSTRACT: Despite being vastly ignored in the literature, coping with topological noise
is an issue of increasing importance, especially as a consequence of the
increasing number and diversity of 3D polygonal models that are captured by
devices of different qualities or synthesized by algorithms of different
stabilities. One approach for matching 3D shapes under topological noise is to
replace the topology-sensitive geodesic distance with distances that are less
sensitive to topological changes. We propose an alternative approach utilising
gradual deflation (or inflation) of the shape volume, of which purpose is to
bring the pair of shapes to be matched to a \emph{comparable} topology before
the search for correspondences. Illustrative experiments using different
datasets demonstrate that as the level of topological noise increases, our
approach outperforms the other methods in the literature.
| no_new_dataset | 0.952882 |
1705.02012 | Tong Wang | Xingdi Yuan, Tong Wang, Caglar Gulcehre, Alessandro Sordoni, Philip
Bachman, Sandeep Subramanian, Saizheng Zhang, Adam Trischler | Machine Comprehension by Text-to-Text Neural Question Generation | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a recurrent neural model that generates natural-language questions
from documents, conditioned on answers. We show how to train the model using a
combination of supervised and reinforcement learning. After teacher forcing for
standard maximum likelihood training, we fine-tune the model using policy
gradient techniques to maximize several rewards that measure question quality.
Most notably, one of these rewards is the performance of a question-answering
system. We motivate question generation as a means to improve the performance
of question answering systems. Our model is trained and evaluated on the recent
question-answering dataset SQuAD.
| [
{
"version": "v1",
"created": "Thu, 4 May 2017 20:58:06 GMT"
},
{
"version": "v2",
"created": "Mon, 15 May 2017 14:47:05 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Yuan",
"Xingdi",
""
],
[
"Wang",
"Tong",
""
],
[
"Gulcehre",
"Caglar",
""
],
[
"Sordoni",
"Alessandro",
""
],
[
"Bachman",
"Philip",
""
],
[
"Subramanian",
"Sandeep",
""
],
[
"Zhang",
"Saizheng",
""
],
[
"Trischler",
"Adam",
""
]
] | TITLE: Machine Comprehension by Text-to-Text Neural Question Generation
ABSTRACT: We propose a recurrent neural model that generates natural-language questions
from documents, conditioned on answers. We show how to train the model using a
combination of supervised and reinforcement learning. After teacher forcing for
standard maximum likelihood training, we fine-tune the model using policy
gradient techniques to maximize several rewards that measure question quality.
Most notably, one of these rewards is the performance of a question-answering
system. We motivate question generation as a means to improve the performance
of question answering systems. Our model is trained and evaluated on the recent
question-answering dataset SQuAD.
| no_new_dataset | 0.927166 |
1705.02519 | Subhabrata Mukherjee | Subhabrata Mukherjee, Hemank Lamba, Gerhard Weikum | Item Recommendation with Evolving User Preferences and Experience | null | null | 10.1109/ICDM.2015.111 | null | cs.AI cs.CL cs.IR cs.SI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current recommender systems exploit user and item similarities by
collaborative filtering. Some advanced methods also consider the temporal
evolution of item ratings as a global background process. However, all prior
methods disregard the individual evolution of a user's experience level and how
this is expressed in the user's writing in a review community. In this paper,
we model the joint evolution of user experience, interest in specific item
facets, writing style, and rating behavior. This way we can generate individual
recommendations that take into account the user's maturity level (e.g.,
recommending art movies rather than blockbusters for a cinematography expert).
As only item ratings and review texts are observables, we capture the user's
experience and interests in a latent model learned from her reviews, vocabulary
and writing style. We develop a generative HMM-LDA model to trace user
evolution, where the Hidden Markov Model (HMM) traces her latent experience
progressing over time -- with solely user reviews and ratings as observables
over time. The facets of a user's interest are drawn from a Latent Dirichlet
Allocation (LDA) model derived from her reviews, as a function of her (again
latent) experience level. In experiments with five real-world datasets, we show
that our model improves the rating prediction over state-of-the-art baselines,
by a substantial margin. We also show, in a use-case study, that our model
performs well in the assessment of user experience levels.
| [
{
"version": "v1",
"created": "Sat, 6 May 2017 19:22:41 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Mukherjee",
"Subhabrata",
""
],
[
"Lamba",
"Hemank",
""
],
[
"Weikum",
"Gerhard",
""
]
] | TITLE: Item Recommendation with Evolving User Preferences and Experience
ABSTRACT: Current recommender systems exploit user and item similarities by
collaborative filtering. Some advanced methods also consider the temporal
evolution of item ratings as a global background process. However, all prior
methods disregard the individual evolution of a user's experience level and how
this is expressed in the user's writing in a review community. In this paper,
we model the joint evolution of user experience, interest in specific item
facets, writing style, and rating behavior. This way we can generate individual
recommendations that take into account the user's maturity level (e.g.,
recommending art movies rather than blockbusters for a cinematography expert).
As only item ratings and review texts are observables, we capture the user's
experience and interests in a latent model learned from her reviews, vocabulary
and writing style. We develop a generative HMM-LDA model to trace user
evolution, where the Hidden Markov Model (HMM) traces her latent experience
progressing over time -- with solely user reviews and ratings as observables
over time. The facets of a user's interest are drawn from a Latent Dirichlet
Allocation (LDA) model derived from her reviews, as a function of her (again
latent) experience level. In experiments with five real-world datasets, we show
that our model improves the rating prediction over state-of-the-art baselines,
by a substantial margin. We also show, in a use-case study, that our model
performs well in the assessment of user experience levels.
| no_new_dataset | 0.958847 |
1705.03551 | Mandar Joshi | Mandar Joshi, Eunsol Choi, Daniel S. Weld, Luke Zettlemoyer | TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for
Reading Comprehension | Added references, fixed typos, minor baseline update | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present TriviaQA, a challenging reading comprehension dataset containing
over 650K question-answer-evidence triples. TriviaQA includes 95K
question-answer pairs authored by trivia enthusiasts and independently gathered
evidence documents, six per question on average, that provide high quality
distant supervision for answering the questions. We show that, in comparison to
other recently introduced large-scale datasets, TriviaQA (1) has relatively
complex, compositional questions, (2) has considerable syntactic and lexical
variability between questions and corresponding answer-evidence sentences, and
(3) requires more cross sentence reasoning to find answers. We also present two
baseline algorithms: a feature-based classifier and a state-of-the-art neural
network, that performs well on SQuAD reading comprehension. Neither approach
comes close to human performance (23% and 40% vs. 80%), suggesting that
TriviaQA is a challenging testbed that is worth significant future study. Data
and code available at -- http://nlp.cs.washington.edu/triviaqa/
| [
{
"version": "v1",
"created": "Tue, 9 May 2017 21:35:07 GMT"
},
{
"version": "v2",
"created": "Sat, 13 May 2017 21:12:37 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Joshi",
"Mandar",
""
],
[
"Choi",
"Eunsol",
""
],
[
"Weld",
"Daniel S.",
""
],
[
"Zettlemoyer",
"Luke",
""
]
] | TITLE: TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for
Reading Comprehension
ABSTRACT: We present TriviaQA, a challenging reading comprehension dataset containing
over 650K question-answer-evidence triples. TriviaQA includes 95K
question-answer pairs authored by trivia enthusiasts and independently gathered
evidence documents, six per question on average, that provide high quality
distant supervision for answering the questions. We show that, in comparison to
other recently introduced large-scale datasets, TriviaQA (1) has relatively
complex, compositional questions, (2) has considerable syntactic and lexical
variability between questions and corresponding answer-evidence sentences, and
(3) requires more cross sentence reasoning to find answers. We also present two
baseline algorithms: a feature-based classifier and a state-of-the-art neural
network, that performs well on SQuAD reading comprehension. Neither approach
comes close to human performance (23% and 40% vs. 80%), suggesting that
TriviaQA is a challenging testbed that is worth significant future study. Data
and code available at -- http://nlp.cs.washington.edu/triviaqa/
| new_dataset | 0.956309 |
1705.04803 | Bhaskar Mitra | Federico Nanni, Bhaskar Mitra, Matt Magnusson and Laura Dietz | Benchmark for Complex Answer Retrieval | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Retrieving paragraphs to populate a Wikipedia article is a challenging task.
The new TREC Complex Answer Retrieval (TREC CAR) track introduces a
comprehensive dataset that targets this retrieval scenario. We present early
results from a variety of approaches -- from standard information retrieval
methods (e.g., tf-idf) to complex systems that using query expansion using
knowledge bases and deep neural networks. The goal is to offer future
participants of this track an overview of some promising approaches to tackle
this problem.
| [
{
"version": "v1",
"created": "Sat, 13 May 2017 09:06:52 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Nanni",
"Federico",
""
],
[
"Mitra",
"Bhaskar",
""
],
[
"Magnusson",
"Matt",
""
],
[
"Dietz",
"Laura",
""
]
] | TITLE: Benchmark for Complex Answer Retrieval
ABSTRACT: Retrieving paragraphs to populate a Wikipedia article is a challenging task.
The new TREC Complex Answer Retrieval (TREC CAR) track introduces a
comprehensive dataset that targets this retrieval scenario. We present early
results from a variety of approaches -- from standard information retrieval
methods (e.g., tf-idf) to complex systems that using query expansion using
knowledge bases and deep neural networks. The goal is to offer future
participants of this track an overview of some promising approaches to tackle
this problem.
| new_dataset | 0.957833 |
1705.04828 | Yiluan Guo | Yiluan Guo, Hossein Nejati, Ngai-Man Cheung | Deep neural networks on graph signals for brain imaging analysis | Accepted by ICIP 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Brain imaging data such as EEG or MEG are high-dimensional spatiotemporal
data often degraded by complex, non-Gaussian noise. For reliable analysis of
brain imaging data, it is important to extract discriminative, low-dimensional
intrinsic representation of the recorded data. This work proposes a new method
to learn the low-dimensional representations from the noise-degraded
measurements. In particular, our work proposes a new deep neural network design
that integrates graph information such as brain connectivity with
fully-connected layers. Our work leverages efficient graph filter design using
Chebyshev polynomial and recent work on convolutional nets on graph-structured
data. Our approach exploits graph structure as the prior side information,
localized graph filter for feature extraction and neural networks for high
capacity learning. Experiments on real MEG datasets show that our approach can
extract more discriminative representations, leading to improved accuracy in a
supervised classification task.
| [
{
"version": "v1",
"created": "Sat, 13 May 2017 13:50:47 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Guo",
"Yiluan",
""
],
[
"Nejati",
"Hossein",
""
],
[
"Cheung",
"Ngai-Man",
""
]
] | TITLE: Deep neural networks on graph signals for brain imaging analysis
ABSTRACT: Brain imaging data such as EEG or MEG are high-dimensional spatiotemporal
data often degraded by complex, non-Gaussian noise. For reliable analysis of
brain imaging data, it is important to extract discriminative, low-dimensional
intrinsic representation of the recorded data. This work proposes a new method
to learn the low-dimensional representations from the noise-degraded
measurements. In particular, our work proposes a new deep neural network design
that integrates graph information such as brain connectivity with
fully-connected layers. Our work leverages efficient graph filter design using
Chebyshev polynomial and recent work on convolutional nets on graph-structured
data. Our approach exploits graph structure as the prior side information,
localized graph filter for feature extraction and neural networks for high
capacity learning. Experiments on real MEG datasets show that our approach can
extract more discriminative representations, leading to improved accuracy in a
supervised classification task.
| no_new_dataset | 0.954732 |
1705.04892 | Jimmy Lin | Jinfeng Rao, Ferhan Ture, Hua He, Oliver Jojic, and Jimmy Lin | Talking to Your TV: Context-Aware Voice Search with Hierarchical
Recurrent Neural Networks | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We tackle the novel problem of navigational voice queries posed against an
entertainment system, where viewers interact with a voice-enabled remote
controller to specify the program to watch. This is a difficult problem for
several reasons: such queries are short, even shorter than comparable voice
queries in other domains, which offers fewer opportunities for deciphering user
intent. Furthermore, ambiguity is exacerbated by underlying speech recognition
errors. We address these challenges by integrating word- and character-level
representations of the queries and by modeling voice search sessions to capture
the contextual dependencies in query sequences. Both are accomplished with a
probabilistic framework in which recurrent and feedforward neural network
modules are organized in a hierarchical manner. From a raw dataset of 32M voice
queries from 2.5M viewers on the Comcast Xfinity X1 entertainment system, we
extracted data to train and test our models. We demonstrate the benefits of our
hybrid representation and context-aware model, which significantly outperforms
models without context as well as the current deployed product.
| [
{
"version": "v1",
"created": "Sat, 13 May 2017 22:24:26 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Rao",
"Jinfeng",
""
],
[
"Ture",
"Ferhan",
""
],
[
"He",
"Hua",
""
],
[
"Jojic",
"Oliver",
""
],
[
"Lin",
"Jimmy",
""
]
] | TITLE: Talking to Your TV: Context-Aware Voice Search with Hierarchical
Recurrent Neural Networks
ABSTRACT: We tackle the novel problem of navigational voice queries posed against an
entertainment system, where viewers interact with a voice-enabled remote
controller to specify the program to watch. This is a difficult problem for
several reasons: such queries are short, even shorter than comparable voice
queries in other domains, which offers fewer opportunities for deciphering user
intent. Furthermore, ambiguity is exacerbated by underlying speech recognition
errors. We address these challenges by integrating word- and character-level
representations of the queries and by modeling voice search sessions to capture
the contextual dependencies in query sequences. Both are accomplished with a
probabilistic framework in which recurrent and feedforward neural network
modules are organized in a hierarchical manner. From a raw dataset of 32M voice
queries from 2.5M viewers on the Comcast Xfinity X1 entertainment system, we
extracted data to train and test our models. We demonstrate the benefits of our
hybrid representation and context-aware model, which significantly outperforms
models without context as well as the current deployed product.
| no_new_dataset | 0.810779 |
1705.04916 | Suryansh Kumar | Suryansh Kumar, Yuchao Dai, Hongdong Li | Spatial-Temporal Union of Subspaces for Multi-body Non-rigid
Structure-from-Motion | Author version of this paper has been accepted by Pattern Recognition
Journal in the special issue on Articulated Motion and Deformable Objects.
This work was originally submitted to ACCV 16 conference on 27th May 2016 for
review | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Non-rigid structure-from-motion (NRSfM) has so far been mostly studied for
recovering 3D structure of a single non-rigid/deforming object. To handle the
real world challenging multiple deforming objects scenarios, existing methods
either pre-segment different objects in the scene or treat multiple non-rigid
objects as a whole to obtain the 3D non-rigid reconstruction. However, these
methods fail to exploit the inherent structure in the problem as the solution
of segmentation and the solution of reconstruction could not benefit each
other. In this paper, we propose a unified framework to jointly segment and
reconstruct multiple non-rigid objects. To compactly represent complex
multi-body non-rigid scenes, we propose to exploit the structure of the scenes
along both temporal direction and spatial direction, thus achieving a
spatio-temporal representation. Specifically, we represent the 3D non-rigid
deformations as lying in a union of subspaces along the temporal direction and
represent the 3D trajectories as lying in the union of subspaces along the
spatial direction. This spatio-temporal representation not only provides
competitive 3D reconstruction but also outputs robust segmentation of multiple
non-rigid objects. The resultant optimization problem is solved efficiently
using the Alternating Direction Method of Multipliers (ADMM). Extensive
experimental results on both synthetic and real multi-body NRSfM datasets
demonstrate the superior performance of our proposed framework compared with
the state-of-the-art methods.
| [
{
"version": "v1",
"created": "Sun, 14 May 2017 05:59:51 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Kumar",
"Suryansh",
""
],
[
"Dai",
"Yuchao",
""
],
[
"Li",
"Hongdong",
""
]
] | TITLE: Spatial-Temporal Union of Subspaces for Multi-body Non-rigid
Structure-from-Motion
ABSTRACT: Non-rigid structure-from-motion (NRSfM) has so far been mostly studied for
recovering 3D structure of a single non-rigid/deforming object. To handle the
real world challenging multiple deforming objects scenarios, existing methods
either pre-segment different objects in the scene or treat multiple non-rigid
objects as a whole to obtain the 3D non-rigid reconstruction. However, these
methods fail to exploit the inherent structure in the problem as the solution
of segmentation and the solution of reconstruction could not benefit each
other. In this paper, we propose a unified framework to jointly segment and
reconstruct multiple non-rigid objects. To compactly represent complex
multi-body non-rigid scenes, we propose to exploit the structure of the scenes
along both temporal direction and spatial direction, thus achieving a
spatio-temporal representation. Specifically, we represent the 3D non-rigid
deformations as lying in a union of subspaces along the temporal direction and
represent the 3D trajectories as lying in the union of subspaces along the
spatial direction. This spatio-temporal representation not only provides
competitive 3D reconstruction but also outputs robust segmentation of multiple
non-rigid objects. The resultant optimization problem is solved efficiently
using the Alternating Direction Method of Multipliers (ADMM). Extensive
experimental results on both synthetic and real multi-body NRSfM datasets
demonstrate the superior performance of our proposed framework compared with
the state-of-the-art methods.
| no_new_dataset | 0.94699 |
1705.04932 | Shuchang Zhou | Shuchang Zhou, Taihong Xiao, Yi Yang, Dieqiao Feng, Qinyao He, Weiran
He | GeneGAN: Learning Object Transfiguration and Attribute Subspace from
Unpaired Data | Github: https://github.com/Prinsphield/GeneGAN | null | null | null | cs.CV cs.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Object Transfiguration replaces an object in an image with another object
from a second image. For example it can perform tasks like "putting exactly
those eyeglasses from image A on the nose of the person in image B". Usage of
exemplar images allows more precise specification of desired modifications and
improves the diversity of conditional image generation. However, previous
methods that rely on feature space operations, require paired data and/or
appearance models for training or disentangling objects from background. In
this work, we propose a model that can learn object transfiguration from two
unpaired sets of images: one set containing images that "have" that kind of
object, and the other set being the opposite, with the mild constraint that the
objects be located approximately at the same place. For example, the training
data can be one set of reference face images that have eyeglasses, and another
set of images that have not, both of which spatially aligned by face landmarks.
Despite the weak 0/1 labels, our model can learn an "eyeglasses" subspace that
contain multiple representatives of different types of glasses. Consequently,
we can perform fine-grained control of generated images, like swapping the
glasses in two images by swapping the projected components in the "eyeglasses"
subspace, to create novel images of people wearing eyeglasses.
Overall, our deterministic generative model learns disentangled attribute
subspaces from weakly labeled data by adversarial training. Experiments on
CelebA and Multi-PIE datasets validate the effectiveness of the proposed model
on real world data, in generating images with specified eyeglasses, smiling,
hair styles, and lighting conditions etc. The code is available online.
| [
{
"version": "v1",
"created": "Sun, 14 May 2017 08:59:36 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Zhou",
"Shuchang",
""
],
[
"Xiao",
"Taihong",
""
],
[
"Yang",
"Yi",
""
],
[
"Feng",
"Dieqiao",
""
],
[
"He",
"Qinyao",
""
],
[
"He",
"Weiran",
""
]
] | TITLE: GeneGAN: Learning Object Transfiguration and Attribute Subspace from
Unpaired Data
ABSTRACT: Object Transfiguration replaces an object in an image with another object
from a second image. For example it can perform tasks like "putting exactly
those eyeglasses from image A on the nose of the person in image B". Usage of
exemplar images allows more precise specification of desired modifications and
improves the diversity of conditional image generation. However, previous
methods that rely on feature space operations, require paired data and/or
appearance models for training or disentangling objects from background. In
this work, we propose a model that can learn object transfiguration from two
unpaired sets of images: one set containing images that "have" that kind of
object, and the other set being the opposite, with the mild constraint that the
objects be located approximately at the same place. For example, the training
data can be one set of reference face images that have eyeglasses, and another
set of images that have not, both of which spatially aligned by face landmarks.
Despite the weak 0/1 labels, our model can learn an "eyeglasses" subspace that
contain multiple representatives of different types of glasses. Consequently,
we can perform fine-grained control of generated images, like swapping the
glasses in two images by swapping the projected components in the "eyeglasses"
subspace, to create novel images of people wearing eyeglasses.
Overall, our deterministic generative model learns disentangled attribute
subspaces from weakly labeled data by adversarial training. Experiments on
CelebA and Multi-PIE datasets validate the effectiveness of the proposed model
on real world data, in generating images with specified eyeglasses, smiling,
hair styles, and lighting conditions etc. The code is available online.
| no_new_dataset | 0.940844 |
1705.04964 | Balint Daroczy | B\'alint Zolt\'an Dar\'oczy | Machine learning methods for multimedia information retrieval | doctoral thesis, 2016 | null | 10.15476/ELTE.2016.086 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this thesis we examined several multimodal feature extraction and learning
methods for retrieval and classification purposes. We reread briefly some
theoretical results of learning in Section 2 and reviewed several generative
and discriminative models in Section 3 while we described the similarity kernel
in Section 4. We examined different aspects of the multimodal image retrieval
and classification in Section 5 and suggested methods for identifying quality
assessments of Web documents in Section 6. In our last problem we proposed
similarity kernel for time-series based classification. The experiments were
carried over publicly available datasets and source codes for the most
essential parts are either open source or released. Since the used similarity
graphs (Section 4.2) are greatly constrained for computational purposes, we
would like to continue work with more complex, evolving and capable graphs and
apply for different problems such as capturing the rapid change in the
distribution (e.g. session based recommendation) or complex graphs of the
literature work. The similarity kernel with the proper metrics reaches and in
many cases improves over the state-of-the-art. Hence we may conclude generative
models based on instance similarities with multiple modes is a generally
applicable model for classification and regression tasks ranging over various
domains, including but not limited to the ones presented in this thesis. More
generally, the Fisher kernel is not only unique in many ways but one of the
most powerful kernel functions. Therefore we may exploit the Fisher kernel in
the future over widely used generative models, such as Boltzmann Machines
[Hinton et al., 1984], a particular subset, the Restricted Boltzmann Machines
and Deep Belief Networks [Hinton et al., 2006]), Latent Dirichlet Allocation
[Blei et al., 2003] or Hidden Markov Models [Baum and Petrie, 1966] to name a
few.
| [
{
"version": "v1",
"created": "Sun, 14 May 2017 14:10:22 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Daróczy",
"Bálint Zoltán",
""
]
] | TITLE: Machine learning methods for multimedia information retrieval
ABSTRACT: In this thesis we examined several multimodal feature extraction and learning
methods for retrieval and classification purposes. We reread briefly some
theoretical results of learning in Section 2 and reviewed several generative
and discriminative models in Section 3 while we described the similarity kernel
in Section 4. We examined different aspects of the multimodal image retrieval
and classification in Section 5 and suggested methods for identifying quality
assessments of Web documents in Section 6. In our last problem we proposed
similarity kernel for time-series based classification. The experiments were
carried over publicly available datasets and source codes for the most
essential parts are either open source or released. Since the used similarity
graphs (Section 4.2) are greatly constrained for computational purposes, we
would like to continue work with more complex, evolving and capable graphs and
apply for different problems such as capturing the rapid change in the
distribution (e.g. session based recommendation) or complex graphs of the
literature work. The similarity kernel with the proper metrics reaches and in
many cases improves over the state-of-the-art. Hence we may conclude generative
models based on instance similarities with multiple modes is a generally
applicable model for classification and regression tasks ranging over various
domains, including but not limited to the ones presented in this thesis. More
generally, the Fisher kernel is not only unique in many ways but one of the
most powerful kernel functions. Therefore we may exploit the Fisher kernel in
the future over widely used generative models, such as Boltzmann Machines
[Hinton et al., 1984], a particular subset, the Restricted Boltzmann Machines
and Deep Belief Networks [Hinton et al., 2006]), Latent Dirichlet Allocation
[Blei et al., 2003] or Hidden Markov Models [Baum and Petrie, 1966] to name a
few.
| no_new_dataset | 0.948537 |
1705.05040 | Kechen Qin | Lu Wang, Nick Beauchamp, Sarah Shugars, and Kechen Qin | Winning on the Merits: The Joint Effects of Content and Style on Debate
Outcomes | Accepted by TACL, 14 pages | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Debate and deliberation play essential roles in politics and government, but
most models presume that debates are won mainly via superior style or agenda
control. Ideally, however, debates would be won on the merits, as a function of
which side has the stronger arguments. We propose a predictive model of debate
that estimates the effects of linguistic features and the latent persuasive
strengths of different topics, as well as the interactions between the two.
Using a dataset of 118 Oxford-style debates, our model's combination of content
(as latent topics) and style (as linguistic features) allows us to predict
audience-adjudicated winners with 74% accuracy, significantly outperforming
linguistic features alone (66%). Our model finds that winning sides employ
stronger arguments, and allows us to identify the linguistic features
associated with strong or weak arguments.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 00:21:03 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Wang",
"Lu",
""
],
[
"Beauchamp",
"Nick",
""
],
[
"Shugars",
"Sarah",
""
],
[
"Qin",
"Kechen",
""
]
] | TITLE: Winning on the Merits: The Joint Effects of Content and Style on Debate
Outcomes
ABSTRACT: Debate and deliberation play essential roles in politics and government, but
most models presume that debates are won mainly via superior style or agenda
control. Ideally, however, debates would be won on the merits, as a function of
which side has the stronger arguments. We propose a predictive model of debate
that estimates the effects of linguistic features and the latent persuasive
strengths of different topics, as well as the interactions between the two.
Using a dataset of 118 Oxford-style debates, our model's combination of content
(as latent topics) and style (as linguistic features) allows us to predict
audience-adjudicated winners with 74% accuracy, significantly outperforming
linguistic features alone (66%). Our model finds that winning sides employ
stronger arguments, and allows us to identify the linguistic features
associated with strong or weak arguments.
| no_new_dataset | 0.942771 |
1705.05084 | Bolun Cai | Xiaoyi Jia, Xiangmin Xu, Bolun Cai, Kailing Guo | Single Image Super-Resolution Using Multi-Scale Convolutional Neural
Network | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Methods based on convolutional neural network (CNN) have demonstrated
tremendous improvements on single image super-resolution. However, the previous
methods mainly restore images from one single area in the low resolution (LR)
input, which limits the flexibility of models to infer various scales of
details for high resolution (HR) output. Moreover, most of them train a
specific model for each up-scale factor. In this paper, we propose a
multi-scale super resolution (MSSR) network. Our network consists of
multi-scale paths to make the HR inference, which can learn to synthesize
features from different scales. This property helps reconstruct various kinds
of regions in HR images. In addition, only one single model is needed for
multiple up-scale factors, which is more efficient without loss of restoration
quality. Experiments on four public datasets demonstrate that the proposed
method achieved state-of-the-art performance with fast speed.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 06:38:04 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Jia",
"Xiaoyi",
""
],
[
"Xu",
"Xiangmin",
""
],
[
"Cai",
"Bolun",
""
],
[
"Guo",
"Kailing",
""
]
] | TITLE: Single Image Super-Resolution Using Multi-Scale Convolutional Neural
Network
ABSTRACT: Methods based on convolutional neural network (CNN) have demonstrated
tremendous improvements on single image super-resolution. However, the previous
methods mainly restore images from one single area in the low resolution (LR)
input, which limits the flexibility of models to infer various scales of
details for high resolution (HR) output. Moreover, most of them train a
specific model for each up-scale factor. In this paper, we propose a
multi-scale super resolution (MSSR) network. Our network consists of
multi-scale paths to make the HR inference, which can learn to synthesize
features from different scales. This property helps reconstruct various kinds
of regions in HR images. In addition, only one single model is needed for
multiple up-scale factors, which is more efficient without loss of restoration
quality. Experiments on four public datasets demonstrate that the proposed
method achieved state-of-the-art performance with fast speed.
| no_new_dataset | 0.950869 |
1705.05207 | Lianwen Jin | Xuefeng Xiao, Yafeng Yang, Tasweer Ahmad, Lianwen Jin and Tianhai
Chang | Design of a Very Compact CNN Classifier for Online Handwritten Chinese
Character Recognition Using DropWeight and Global Pooling | 5 pages, 2 figures, 2 tables | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Currently, owing to the ubiquity of mobile devices, online handwritten
Chinese character recognition (HCCR) has become one of the suitable choice for
feeding input to cell phones and tablet devices. Over the past few years,
larger and deeper convolutional neural networks (CNNs) have extensively been
employed for improving character recognition performance. However, its
substantial storage requirement is a significant obstacle in deploying such
networks into portable electronic devices. To circumvent this problem, we
propose a novel technique called DropWeight for pruning redundant connections
in the CNN architecture. It is revealed that the proposed method not only
treats streamlined architectures such as AlexNet and VGGNet well but also
exhibits remarkable performance for deep residual network and inception
network. We also demonstrate that global pooling is a better choice for
building very compact online HCCR systems. Experiments were performed on the
ICDAR-2013 online HCCR competition dataset using our proposed network, and it
is found that the proposed approach requires only 0.57 MB for storage, whereas
state-of-the-art CNN-based methods require up to 135 MB; meanwhile the
performance is decreased only by 0.91%.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 13:18:38 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Xiao",
"Xuefeng",
""
],
[
"Yang",
"Yafeng",
""
],
[
"Ahmad",
"Tasweer",
""
],
[
"Jin",
"Lianwen",
""
],
[
"Chang",
"Tianhai",
""
]
] | TITLE: Design of a Very Compact CNN Classifier for Online Handwritten Chinese
Character Recognition Using DropWeight and Global Pooling
ABSTRACT: Currently, owing to the ubiquity of mobile devices, online handwritten
Chinese character recognition (HCCR) has become one of the suitable choice for
feeding input to cell phones and tablet devices. Over the past few years,
larger and deeper convolutional neural networks (CNNs) have extensively been
employed for improving character recognition performance. However, its
substantial storage requirement is a significant obstacle in deploying such
networks into portable electronic devices. To circumvent this problem, we
propose a novel technique called DropWeight for pruning redundant connections
in the CNN architecture. It is revealed that the proposed method not only
treats streamlined architectures such as AlexNet and VGGNet well but also
exhibits remarkable performance for deep residual network and inception
network. We also demonstrate that global pooling is a better choice for
building very compact online HCCR systems. Experiments were performed on the
ICDAR-2013 online HCCR competition dataset using our proposed network, and it
is found that the proposed approach requires only 0.57 MB for storage, whereas
state-of-the-art CNN-based methods require up to 135 MB; meanwhile the
performance is decreased only by 0.91%.
| no_new_dataset | 0.943556 |
1705.05301 | Paschalis Panteleris | Paschalis Panteleris (1) and Antonis Argyros (1 and 2) ((1) Institute
of Computer Science, FORTH, (2) Computer Science Department, University of
Crete) | Back to RGB: 3D tracking of hands and hand-object interactions based on
short-baseline stereo | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a novel solution to the problem of 3D tracking of the articulated
motion of human hand(s), possibly in interaction with other objects. The vast
majority of contemporary relevant work capitalizes on depth information
provided by RGBD cameras. In this work, we show that accurate and efficient 3D
hand tracking is possible, even for the case of RGB stereo. A straightforward
approach for solving the problem based on such input would be to first recover
depth and then apply a state of the art depth-based 3D hand tracking method.
Unfortunately, this does not work well in practice because the stereo-based,
dense 3D reconstruction of hands is far less accurate than the one obtained by
RGBD cameras. Our approach bypasses 3D reconstruction and follows a completely
different route: 3D hand tracking is formulated as an optimization problem
whose solution is the hand configuration that maximizes the color consistency
between the two views of the hand. We demonstrate the applicability of our
method for real time tracking of a single hand, of a hand manipulating an
object and of two interacting hands. The method has been evaluated
quantitatively on standard datasets and in comparison to relevant, state of the
art RGBD-based approaches. The obtained results demonstrate that the proposed
stereo-based method performs equally well to its RGBD-based competitors, and in
some cases, it even outperforms them.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 15:38:56 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Panteleris",
"Paschalis",
"",
"1 and 2"
],
[
"Argyros",
"Antonis",
"",
"1 and 2"
]
] | TITLE: Back to RGB: 3D tracking of hands and hand-object interactions based on
short-baseline stereo
ABSTRACT: We present a novel solution to the problem of 3D tracking of the articulated
motion of human hand(s), possibly in interaction with other objects. The vast
majority of contemporary relevant work capitalizes on depth information
provided by RGBD cameras. In this work, we show that accurate and efficient 3D
hand tracking is possible, even for the case of RGB stereo. A straightforward
approach for solving the problem based on such input would be to first recover
depth and then apply a state of the art depth-based 3D hand tracking method.
Unfortunately, this does not work well in practice because the stereo-based,
dense 3D reconstruction of hands is far less accurate than the one obtained by
RGBD cameras. Our approach bypasses 3D reconstruction and follows a completely
different route: 3D hand tracking is formulated as an optimization problem
whose solution is the hand configuration that maximizes the color consistency
between the two views of the hand. We demonstrate the applicability of our
method for real time tracking of a single hand, of a hand manipulating an
object and of two interacting hands. The method has been evaluated
quantitatively on standard datasets and in comparison to relevant, state of the
art RGBD-based approaches. The obtained results demonstrate that the proposed
stereo-based method performs equally well to its RGBD-based competitors, and in
some cases, it even outperforms them.
| no_new_dataset | 0.941547 |
1705.05347 | Rafael Uetz | Luis Alberto Benthin Sanguino, Rafael Uetz | Software Vulnerability Analysis Using CPE and CVE | null | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we analyze the Common Platform Enumeration (CPE) dictionary
and the Common Vulnerabilities and Exposures (CVE) feeds. These repositories
are widely used in Vulnerability Management Systems (VMSs) to check for known
vulnerabilities in software products. The analysis shows, among other issues, a
lack of synchronization between both datasets that can lead to incorrect
results output by VMSs relying on those datasets. To deal with these problems,
we developed a method that recommends to a user a prioritized list of CPE
identifiers for a given software product. The user can then assign (and, if
necessary, adapt) the most suitable CPE identifier to the software so that
regular (e.g., daily) checks can find known vulnerabilities for this software
in the CVE feeds. Our evaluation of this method shows that this interaction is
indeed necessary because a fully automated CPE assignment is prone to errors
due to the CPE and CVE shortcomings. We implemented an open-source VMS that
employs the proposed method and published it on GitHub.
| [
{
"version": "v1",
"created": "Mon, 15 May 2017 17:33:47 GMT"
}
] | 2017-05-16T00:00:00 | [
[
"Sanguino",
"Luis Alberto Benthin",
""
],
[
"Uetz",
"Rafael",
""
]
] | TITLE: Software Vulnerability Analysis Using CPE and CVE
ABSTRACT: In this paper, we analyze the Common Platform Enumeration (CPE) dictionary
and the Common Vulnerabilities and Exposures (CVE) feeds. These repositories
are widely used in Vulnerability Management Systems (VMSs) to check for known
vulnerabilities in software products. The analysis shows, among other issues, a
lack of synchronization between both datasets that can lead to incorrect
results output by VMSs relying on those datasets. To deal with these problems,
we developed a method that recommends to a user a prioritized list of CPE
identifiers for a given software product. The user can then assign (and, if
necessary, adapt) the most suitable CPE identifier to the software so that
regular (e.g., daily) checks can find known vulnerabilities for this software
in the CVE feeds. Our evaluation of this method shows that this interaction is
indeed necessary because a fully automated CPE assignment is prone to errors
due to the CPE and CVE shortcomings. We implemented an open-source VMS that
employs the proposed method and published it on GitHub.
| no_new_dataset | 0.943608 |
1510.05198 | Jiwei Li | Jiwei Li, Alan Ritter and Dan Jurafsky | Learning multi-faceted representations of individuals from heterogeneous
evidence using neural networks | null | null | null | null | cs.SI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Inferring latent attributes of people online is an important social computing
task, but requires integrating the many heterogeneous sources of information
available on the web. We propose learning individual representations of people
using neural nets to integrate rich linguistic and network evidence gathered
from social media. The algorithm is able to combine diverse cues, such as the
text a person writes, their attributes (e.g. gender, employer, education,
location) and social relations to other people. We show that by integrating
both textual and network evidence, these representations offer improved
performance at four important tasks in social media inference on Twitter:
predicting (1) gender, (2) occupation, (3) location, and (4) friendships for
users. Our approach scales to large datasets and the learned representations
can be used as general features in and have the potential to benefit a large
number of downstream tasks including link prediction, community detection, or
probabilistic reasoning over social networks.
| [
{
"version": "v1",
"created": "Sun, 18 Oct 2015 04:26:08 GMT"
},
{
"version": "v2",
"created": "Fri, 30 Oct 2015 22:45:41 GMT"
},
{
"version": "v3",
"created": "Wed, 22 Feb 2017 06:20:53 GMT"
},
{
"version": "v4",
"created": "Thu, 11 May 2017 20:47:13 GMT"
}
] | 2017-05-15T00:00:00 | [
[
"Li",
"Jiwei",
""
],
[
"Ritter",
"Alan",
""
],
[
"Jurafsky",
"Dan",
""
]
] | TITLE: Learning multi-faceted representations of individuals from heterogeneous
evidence using neural networks
ABSTRACT: Inferring latent attributes of people online is an important social computing
task, but requires integrating the many heterogeneous sources of information
available on the web. We propose learning individual representations of people
using neural nets to integrate rich linguistic and network evidence gathered
from social media. The algorithm is able to combine diverse cues, such as the
text a person writes, their attributes (e.g. gender, employer, education,
location) and social relations to other people. We show that by integrating
both textual and network evidence, these representations offer improved
performance at four important tasks in social media inference on Twitter:
predicting (1) gender, (2) occupation, (3) location, and (4) friendships for
users. Our approach scales to large datasets and the learned representations
can be used as general features in and have the potential to benefit a large
number of downstream tasks including link prediction, community detection, or
probabilistic reasoning over social networks.
| no_new_dataset | 0.947866 |
1511.04646 | Yikang Shen | Yikang Shen, Wenge Rong, Nan Jiang, Baolin Peng, Jie Tang, Zhang Xiong | Word Embedding based Correlation Model for Question/Answer Matching | 8 pages, 2 figures | AAAI (2017) 3511--3517 | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the development of community based question answering (Q&A) services, a
large scale of Q&A archives have been accumulated and are an important
information and knowledge resource on the web. Question and answer matching has
been attached much importance to for its ability to reuse knowledge stored in
these systems: it can be useful in enhancing user experience with recurrent
questions. In this paper, we try to improve the matching accuracy by overcoming
the lexical gap between question and answer pairs. A Word Embedding based
Correlation (WEC) model is proposed by integrating advantages of both the
translation model and word embedding, given a random pair of words, WEC can
score their co-occurrence probability in Q&A pairs and it can also leverage the
continuity and smoothness of continuous space word representation to deal with
new pairs of words that are rare in the training parallel text. An experimental
study on Yahoo! Answers dataset and Baidu Zhidao dataset shows this new
method's promising potential.
| [
{
"version": "v1",
"created": "Sun, 15 Nov 2015 02:59:22 GMT"
},
{
"version": "v2",
"created": "Sat, 26 Nov 2016 02:40:12 GMT"
}
] | 2017-05-15T00:00:00 | [
[
"Shen",
"Yikang",
""
],
[
"Rong",
"Wenge",
""
],
[
"Jiang",
"Nan",
""
],
[
"Peng",
"Baolin",
""
],
[
"Tang",
"Jie",
""
],
[
"Xiong",
"Zhang",
""
]
] | TITLE: Word Embedding based Correlation Model for Question/Answer Matching
ABSTRACT: With the development of community based question answering (Q&A) services, a
large scale of Q&A archives have been accumulated and are an important
information and knowledge resource on the web. Question and answer matching has
been attached much importance to for its ability to reuse knowledge stored in
these systems: it can be useful in enhancing user experience with recurrent
questions. In this paper, we try to improve the matching accuracy by overcoming
the lexical gap between question and answer pairs. A Word Embedding based
Correlation (WEC) model is proposed by integrating advantages of both the
translation model and word embedding, given a random pair of words, WEC can
score their co-occurrence probability in Q&A pairs and it can also leverage the
continuity and smoothness of continuous space word representation to deal with
new pairs of words that are rare in the training parallel text. An experimental
study on Yahoo! Answers dataset and Baidu Zhidao dataset shows this new
method's promising potential.
| no_new_dataset | 0.949201 |
1605.04129 | Maedeh Aghaei | Maedeh Aghaei, Mariella Dimiccoli, Petia Radeva | With Whom Do I Interact? Detecting Social Interactions in Egocentric
Photo-streams | 6 pages, 9 figures, accepted and presented in International
Conference on Pattern Recognition (ICPR 2016) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given a user wearing a low frame rate wearable camera during a day, this work
aims to automatically detect the moments when the user gets engaged into a
social interaction solely by reviewing the automatically captured photos by the
worn camera. The proposed method, inspired by the sociological concept of
F-formation, exploits distance and orientation of the appearing individuals
-with respect to the user- in the scene from a bird-view perspective. As a
result, the interaction pattern over the sequence can be understood as a
two-dimensional time series that corresponds to the temporal evolution of the
distance and orientation features over time. A Long-Short Term Memory-based
Recurrent Neural Network is then trained to classify each time series.
Experimental evaluation over a dataset of 30.000 images has shown promising
results on the proposed method for social interaction detection in egocentric
photo-streams.
| [
{
"version": "v1",
"created": "Fri, 13 May 2016 11:04:28 GMT"
},
{
"version": "v2",
"created": "Fri, 12 May 2017 11:27:50 GMT"
}
] | 2017-05-15T00:00:00 | [
[
"Aghaei",
"Maedeh",
""
],
[
"Dimiccoli",
"Mariella",
""
],
[
"Radeva",
"Petia",
""
]
] | TITLE: With Whom Do I Interact? Detecting Social Interactions in Egocentric
Photo-streams
ABSTRACT: Given a user wearing a low frame rate wearable camera during a day, this work
aims to automatically detect the moments when the user gets engaged into a
social interaction solely by reviewing the automatically captured photos by the
worn camera. The proposed method, inspired by the sociological concept of
F-formation, exploits distance and orientation of the appearing individuals
-with respect to the user- in the scene from a bird-view perspective. As a
result, the interaction pattern over the sequence can be understood as a
two-dimensional time series that corresponds to the temporal evolution of the
distance and orientation features over time. A Long-Short Term Memory-based
Recurrent Neural Network is then trained to classify each time series.
Experimental evaluation over a dataset of 30.000 images has shown promising
results on the proposed method for social interaction detection in egocentric
photo-streams.
| no_new_dataset | 0.707809 |
1606.00915 | Liang-Chieh Chen | Liang-Chieh Chen and George Papandreou and Iasonas Kokkinos and Kevin
Murphy and Alan L. Yuille | DeepLab: Semantic Image Segmentation with Deep Convolutional Nets,
Atrous Convolution, and Fully Connected CRFs | Accepted by TPAMI | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work we address the task of semantic image segmentation with Deep
Learning and make three main contributions that are experimentally shown to
have substantial practical merit. First, we highlight convolution with
upsampled filters, or 'atrous convolution', as a powerful tool in dense
prediction tasks. Atrous convolution allows us to explicitly control the
resolution at which feature responses are computed within Deep Convolutional
Neural Networks. It also allows us to effectively enlarge the field of view of
filters to incorporate larger context without increasing the number of
parameters or the amount of computation. Second, we propose atrous spatial
pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP
probes an incoming convolutional feature layer with filters at multiple
sampling rates and effective fields-of-views, thus capturing objects as well as
image context at multiple scales. Third, we improve the localization of object
boundaries by combining methods from DCNNs and probabilistic graphical models.
The commonly deployed combination of max-pooling and downsampling in DCNNs
achieves invariance but has a toll on localization accuracy. We overcome this
by combining the responses at the final DCNN layer with a fully connected
Conditional Random Field (CRF), which is shown both qualitatively and
quantitatively to improve localization performance. Our proposed "DeepLab"
system sets the new state-of-art at the PASCAL VOC-2012 semantic image
segmentation task, reaching 79.7% mIOU in the test set, and advances the
results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and
Cityscapes. All of our code is made publicly available online.
| [
{
"version": "v1",
"created": "Thu, 2 Jun 2016 21:52:21 GMT"
},
{
"version": "v2",
"created": "Fri, 12 May 2017 03:25:47 GMT"
}
] | 2017-05-15T00:00:00 | [
[
"Chen",
"Liang-Chieh",
""
],
[
"Papandreou",
"George",
""
],
[
"Kokkinos",
"Iasonas",
""
],
[
"Murphy",
"Kevin",
""
],
[
"Yuille",
"Alan L.",
""
]
] | TITLE: DeepLab: Semantic Image Segmentation with Deep Convolutional Nets,
Atrous Convolution, and Fully Connected CRFs
ABSTRACT: In this work we address the task of semantic image segmentation with Deep
Learning and make three main contributions that are experimentally shown to
have substantial practical merit. First, we highlight convolution with
upsampled filters, or 'atrous convolution', as a powerful tool in dense
prediction tasks. Atrous convolution allows us to explicitly control the
resolution at which feature responses are computed within Deep Convolutional
Neural Networks. It also allows us to effectively enlarge the field of view of
filters to incorporate larger context without increasing the number of
parameters or the amount of computation. Second, we propose atrous spatial
pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP
probes an incoming convolutional feature layer with filters at multiple
sampling rates and effective fields-of-views, thus capturing objects as well as
image context at multiple scales. Third, we improve the localization of object
boundaries by combining methods from DCNNs and probabilistic graphical models.
The commonly deployed combination of max-pooling and downsampling in DCNNs
achieves invariance but has a toll on localization accuracy. We overcome this
by combining the responses at the final DCNN layer with a fully connected
Conditional Random Field (CRF), which is shown both qualitatively and
quantitatively to improve localization performance. Our proposed "DeepLab"
system sets the new state-of-art at the PASCAL VOC-2012 semantic image
segmentation task, reaching 79.7% mIOU in the test set, and advances the
results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and
Cityscapes. All of our code is made publicly available online.
| no_new_dataset | 0.954052 |
1611.07661 | Michael Maire | Tsung-Wei Ke, Michael Maire, Stella X. Yu | Multigrid Neural Architectures | updated with ImageNet results; to appear at CVPR 2017 | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a multigrid extension of convolutional neural networks (CNNs).
Rather than manipulating representations living on a single spatial grid, our
network layers operate across scale space, on a pyramid of grids. They consume
multigrid inputs and produce multigrid outputs; convolutional filters
themselves have both within-scale and cross-scale extent. This aspect is
distinct from simple multiscale designs, which only process the input at
different scales. Viewed in terms of information flow, a multigrid network
passes messages across a spatial pyramid. As a consequence, receptive field
size grows exponentially with depth, facilitating rapid integration of context.
Most critically, multigrid structure enables networks to learn internal
attention and dynamic routing mechanisms, and use them to accomplish tasks on
which modern CNNs fail.
Experiments demonstrate wide-ranging performance advantages of multigrid. On
CIFAR and ImageNet classification tasks, flipping from a single grid to
multigrid within the standard CNN paradigm improves accuracy, while being
compute and parameter efficient. Multigrid is independent of other
architectural choices; we show synergy in combination with residual
connections. Multigrid yields dramatic improvement on a synthetic semantic
segmentation dataset. Most strikingly, relatively shallow multigrid networks
can learn to directly perform spatial transformation tasks, where, in contrast,
current CNNs fail. Together, our results suggest that continuous evolution of
features on a multigrid pyramid is a more powerful alternative to existing CNN
designs on a flat grid.
| [
{
"version": "v1",
"created": "Wed, 23 Nov 2016 06:55:53 GMT"
},
{
"version": "v2",
"created": "Thu, 11 May 2017 19:24:33 GMT"
}
] | 2017-05-15T00:00:00 | [
[
"Ke",
"Tsung-Wei",
""
],
[
"Maire",
"Michael",
""
],
[
"Yu",
"Stella X.",
""
]
] | TITLE: Multigrid Neural Architectures
ABSTRACT: We propose a multigrid extension of convolutional neural networks (CNNs).
Rather than manipulating representations living on a single spatial grid, our
network layers operate across scale space, on a pyramid of grids. They consume
multigrid inputs and produce multigrid outputs; convolutional filters
themselves have both within-scale and cross-scale extent. This aspect is
distinct from simple multiscale designs, which only process the input at
different scales. Viewed in terms of information flow, a multigrid network
passes messages across a spatial pyramid. As a consequence, receptive field
size grows exponentially with depth, facilitating rapid integration of context.
Most critically, multigrid structure enables networks to learn internal
attention and dynamic routing mechanisms, and use them to accomplish tasks on
which modern CNNs fail.
Experiments demonstrate wide-ranging performance advantages of multigrid. On
CIFAR and ImageNet classification tasks, flipping from a single grid to
multigrid within the standard CNN paradigm improves accuracy, while being
compute and parameter efficient. Multigrid is independent of other
architectural choices; we show synergy in combination with residual
connections. Multigrid yields dramatic improvement on a synthetic semantic
segmentation dataset. Most strikingly, relatively shallow multigrid networks
can learn to directly perform spatial transformation tasks, where, in contrast,
current CNNs fail. Together, our results suggest that continuous evolution of
features on a multigrid pyramid is a more powerful alternative to existing CNN
designs on a flat grid.
| no_new_dataset | 0.950365 |
1703.01790 | Maedeh Aghaei | Maedeh Aghaei, Mariella Dimiccoli, Petia Radeva | All the people around me: face discovery in egocentric photo-streams | 5 pages, 3 figures, accepted in IEEE International Conference on
Image Processing (ICIP 2017) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given an unconstrained stream of images captured by a wearable photo-camera
(2fpm), we propose an unsupervised bottom-up approach for automatic clustering
appearing faces into the individual identities present in these data. The
problem is challenging since images are acquired under real world conditions;
hence the visible appearance of the people in the images undergoes intensive
variations. Our proposed pipeline consists of first arranging the photo-stream
into events, later, localizing the appearance of multiple people in them, and
finally, grouping various appearances of the same person across different
events. Experimental results performed on a dataset acquired by wearing a
photo-camera during one month, demonstrate the effectiveness of the proposed
approach for the considered purpose.
| [
{
"version": "v1",
"created": "Mon, 6 Mar 2017 09:50:39 GMT"
},
{
"version": "v2",
"created": "Fri, 12 May 2017 11:29:38 GMT"
}
] | 2017-05-15T00:00:00 | [
[
"Aghaei",
"Maedeh",
""
],
[
"Dimiccoli",
"Mariella",
""
],
[
"Radeva",
"Petia",
""
]
] | TITLE: All the people around me: face discovery in egocentric photo-streams
ABSTRACT: Given an unconstrained stream of images captured by a wearable photo-camera
(2fpm), we propose an unsupervised bottom-up approach for automatic clustering
appearing faces into the individual identities present in these data. The
problem is challenging since images are acquired under real world conditions;
hence the visible appearance of the people in the images undergoes intensive
variations. Our proposed pipeline consists of first arranging the photo-stream
into events, later, localizing the appearance of multiple people in them, and
finally, grouping various appearances of the same person across different
events. Experimental results performed on a dataset acquired by wearing a
photo-camera during one month, demonstrate the effectiveness of the proposed
approach for the considered purpose.
| no_new_dataset | 0.917967 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.