test_response.xml

<?xml version="1.0" encoding="UTF-8"?>

<rdf:RDF
 xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
 xmlns="http://purl.org/rss/1.0/"
 xmlns:content="http://purl.org/rss/1.0/modules/content/"
 xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/"
 xmlns:dc="http://purl.org/dc/elements/1.1/"
 xmlns:syn="http://purl.org/rss/1.0/modules/syndication/"
 xmlns:admin="http://webns.net/mvcb/"
>

<channel rdf:about="http://arxiv.org/">
<title>cs.CV updates on arXiv.org</title>
<link>http://arxiv.org/</link>
<description rdf:parseType="Literal">Computer Science -- Computer Vision and Pattern Recognition (cs.CV) updates on the arXiv.org e-print archive</description>
<dc:language>en-us</dc:language>
<dc:date>2016-05-26T20:30:00-05:00</dc:date>
<dc:publisher>help@arxiv.org</dc:publisher>
<dc:subject>Computer Science -- Computer Vision and Pattern Recognition</dc:subject>
<syn:updateBase>1901-01-01T00:00+00:00</syn:updateBase>
<syn:updateFrequency>1</syn:updateFrequency>
<syn:updatePeriod>daily</syn:updatePeriod>
<items>
 <rdf:Seq>
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08068" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08104" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08110" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08125" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08140" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08151" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08153" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08154" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08163" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08179" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08247" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08283" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08313" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08323" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08350" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08359" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08397" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08401" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.08412" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1602.02660" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1602.02720" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1602.05531" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1604.07666" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1604.08806" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.05579" />
  <rdf:li rdf:resource="http://arxiv.org/abs/1605.07912" />
 </rdf:Seq>
</items>
<image rdf:resource="http://arxiv.org/icons/sfx.gif" />
</channel>
<image rdf:about="http://arxiv.org/icons/sfx.gif">
<title>arXiv.org</title>
<url>http://arxiv.org/icons/sfx.gif</url>
<link>http://arxiv.org/</link>
</image>
<item rdf:about="http://arxiv.org/abs/1605.08068">
<title>Real-Time Human Motion Capture with Multiple Depth Cameras. (arXiv:1605.08068v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08068</link>
<description rdf:parseType="Literal">&lt;p&gt;Commonly used human motion capture systems require intrusive attachment of
markers that are visually tracked with multiple cameras. In this work we
present an efficient and inexpensive solution to markerless motion capture
using only a few Kinect sensors. Unlike the previous work on 3d pose estimation
using a single depth camera, we relax constraints on the camera location and do
not assume a co-operative user. We apply recent image segmentation techniques
to depth images and use curriculum learning to train our system on purely
synthetic data. Our method accurately localizes body parts without requiring an
explicit shape model. The body joint locations are then recovered by combining
evidence from multiple views in real-time. We also introduce a dataset of ~6
million synthetic depth frames for pose estimation from multiple cameras and
exceed state-of-the-art results on the Berkeley MHAD dataset.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Shafaei_A/0/1/0/all/0/1&quot;&gt;Alireza Shafaei&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Little_J/0/1/0/all/0/1&quot;&gt;James J. Little&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08104">
<title>Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning. (arXiv:1605.08104v1 [cs.LG])</title>
<link>http://arxiv.org/abs/1605.08104</link>
<description rdf:parseType="Literal">&lt;p&gt;While great strides have been made in using deep learning algorithms to solve
supervised learning tasks, the problem of unsupervised learning - leveraging
unlabeled examples to learn about the structure of a domain - remains a
difficult unsolved challenge. Here, we explore prediction of future frames in a
video sequence as an unsupervised learning rule for learning about the
structure of the visual world. We describe a predictive neural network
(&quot;PredNet&quot;) architecture that is inspired by the concept of &quot;predictive coding&quot;
from the neuroscience literature. These networks learn to predict future frames
in a video sequence, with each layer in the network making local predictions
and only forwarding deviations from those predictions to subsequent network
layers. We show that these networks are able to robustly learn to predict the
movement of synthetic (rendered) objects, and that in doing so, the networks
learn internal representations that are useful for decoding latent object
parameters (e.g. pose) that support object recognition with fewer training
views. We also show that these networks can scale to complex natural image
streams (car-mounted camera videos), capturing key aspects of both egocentric
movement and the movement of objects in the visual scene, and generalizing
across video datasets. These results suggest that prediction represents a
powerful framework for unsupervised learning, allowing for implicit learning of
object and scene structure.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Lotter_W/0/1/0/all/0/1&quot;&gt;William Lotter&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Kreiman_G/0/1/0/all/0/1&quot;&gt;Gabriel Kreiman&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Cox_D/0/1/0/all/0/1&quot;&gt;David Cox&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08110">
<title>Video Summarization with Long Short-term Memory. (arXiv:1605.08110v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08110</link>
<description rdf:parseType="Literal">&lt;p&gt;We propose a novel supervised learning technique for summarizing videos by
automatically selecting keyframes or key subshots. Casting the problem as a
structured prediction problem on sequential data, our main idea is to use Long
Short-Term Memory (LSTM), a special type of recurrent neural networks to model
the variable-range dependencies entailed in the task of video summarization.
Our learning models attain the state-of-the-art results on two benchmark video
datasets. Detailed analysis justifies the design of the models. In particular,
we show that it is crucial to take into consideration the sequential structures
in videos and model them. Besides advances in modeling techniques, we introduce
techniques to address the need of a large number of annotated data for training
complex learning models. There, our main idea is to exploit the existence of
auxiliary annotated video datasets, albeit heterogeneous in visual styles and
contents. Specifically, we show domain adaptation techniques can improve
summarization by reducing the discrepancies in statistical properties across
those datasets.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Zhang_K/0/1/0/all/0/1&quot;&gt;Ke Zhang&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Chao_W/0/1/0/all/0/1&quot;&gt;Wei-Lun Chao&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Sha_F/0/1/0/all/0/1&quot;&gt;Fei Sha&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Grauman_K/0/1/0/all/0/1&quot;&gt;Kristen Grauman&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08125">
<title>Automatic Action Annotation in Weakly Labeled Videos. (arXiv:1605.08125v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08125</link>
<description rdf:parseType="Literal">&lt;p&gt;Manual spatio-temporal annotation of human action in videos is laborious,
requires several annotators and contains human biases. In this paper, we
present a weakly supervised approach to automatically obtain spatio-temporal
annotations of an actor in action videos. We first obtain a large number of
action proposals in each video. To capture a few most representative action
proposals in each video and evade processing thousands of them, we rank them
using optical flow and saliency in a 3D-MRF based framework and select a few
proposals using MAP based proposal subset selection method. We demonstrate that
this ranking preserves the high quality action proposals. Several such
proposals are generated for each video of the same action. Our next challenge
is to iteratively select one proposal from each video so that all proposals are
globally consistent. We formulate this as Generalized Maximum Clique Graph
problem using shape, global and fine grained similarity of proposals across the
videos. The output of our method is the most action representative proposals
from each video. Our method can also annotate multiple instances of the same
action in a video. We have validated our approach on three challenging action
datasets: UCF Sport, sub-JHMDB and THUMOS&apos;13 and have obtained promising
results compared to several baseline methods. Moreover, on UCF Sports, we
demonstrate that action classifiers trained on these automatically obtained
spatio-temporal annotations have comparable performance to the classifiers
trained on ground truth annotation.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Sultani_W/0/1/0/all/0/1&quot;&gt;Waqas Sultani&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Shah_M/0/1/0/all/0/1&quot;&gt;Mubarak Shah&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08140">
<title>Temporal attention filters for human activity recognition in videos. (arXiv:1605.08140v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08140</link>
<description rdf:parseType="Literal">&lt;p&gt;In this paper, we newly introduce the concept of temporal attention filters,
and describe how they can be used for human activity recognition from videos.
Many high-level activities are often composed of multiple temporal parts (e.g.,
sub-events) with different duration/speed, and our objective is to make the
model explicitly consider such temporal structure using multiple temporal
filters. Our attention filters are designed to be fully differentiable,
allowing end-of-end training of the temporal filters together with the
underlying frame-based or segment-based convolutional neural network
architectures. The paper not only presents an approach of learning optimal
static temporal attention filters to be shared across different videos, but
also describes an approach of dynamically adjusting attention filters per
testing video using recurrent long short-term memory networks (LSTMs). We
experimentally confirm that the proposed concept of temporal attention filters
benefits the activity recognition tasks by capturing the temporal structure in
videos.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Piergiovanni_A/0/1/0/all/0/1&quot;&gt;AJ Piergiovanni&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Fan_C/0/1/0/all/0/1&quot;&gt;Chenyou Fan&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Ryoo_M/0/1/0/all/0/1&quot;&gt;Michael S. Ryoo&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08151">
<title>Predicting Visual Exemplars of Unseen Classes for Zero-Shot Learning. (arXiv:1605.08151v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08151</link>
<description rdf:parseType="Literal">&lt;p&gt;Leveraging class semantic descriptions and examples of known objects,
zero-shot learning makes it possible to train a recognition model for an object
class whose examples are not available. In this paper, we propose a novel
zero-shot learning model that takes advantage of clustering structures in the
semantic embedding space. The key idea is to impose the structural constraint
that semantic representations must be predictive of the locations of its
corresponding visual exemplars. To this end, this reduces to training multiple
kernel-based regressors from semantic representation-exemplar pairs from
labeled data of the seen object categories. Despite its simplicity, our
approach significantly outperforms existing zero-shot learning methods in three
out of four benchmark datasets, including the ImageNet dataset with more than
20,000 unseen categories.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Changpinyo_S/0/1/0/all/0/1&quot;&gt;Soravit Changpinyo&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Chao_W/0/1/0/all/0/1&quot;&gt;Wei-Lun Chao&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Sha_F/0/1/0/all/0/1&quot;&gt;Fei Sha&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08153">
<title>DeepMovie: Using Optical Flow and Deep Neural Networks to Stylize Movies. (arXiv:1605.08153v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08153</link>
<description rdf:parseType="Literal">&lt;p&gt;A recent paper by Gatys et al. describes a method for rendering an image in
the style of another image. First, they use convolutional neural network
features to build a statistical model for the style of an image. Then they
create a new image with the content of one image but the style statistics of
another image. Here, we extend this method to render a movie in a given
artistic style. The naive solution that independently renders each frame
produces poor results because the features of the style move substantially from
one frame to the next. The other naive method that initializes the optimization
for the next frame using the rendered version of the previous frame also
produces poor results because the features of the texture stay fixed relative
to the frame of the movie instead of moving with objects in the scene. The main
contribution of this paper is to use optical flow to initialize the style
transfer optimization so that the texture features move with the objects in the
video. Finally, we suggest a method to incorporate optical flow explicitly into
the cost function.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Anderson_A/0/1/0/all/0/1&quot;&gt;Alexander G. Anderson&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Berg_C/0/1/0/all/0/1&quot;&gt;Cory P. Berg&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Mossing_D/0/1/0/all/0/1&quot;&gt;Daniel P. Mossing&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Olshausen_B/0/1/0/all/0/1&quot;&gt;Bruno A. Olshausen&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08154">
<title>A single scale retinex based method for palm vein extraction. (arXiv:1605.08154v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08154</link>
<description rdf:parseType="Literal">&lt;p&gt;Palm vein recognition is a novel biometric identification technology. But how
to gain a better vein extraction result from the raw palm image is still a
challenging problem, especially when the raw data collection has the problem of
asymmetric illumination. This paper proposes a method based on single scale
Retinex algorithm to extract palm vein image when strong shadow presents due to
asymmetric illumination and uneven geometry of the palm. We test our method on
a multispectral palm image. The experimental result shows that the proposed
method is robust to the influence of illumination angle and shadow. Compared to
the traditional extraction methods, the proposed method can obtain palm vein
lines with better visualization performance (the contrast ratio increases by
18.4%, entropy increases by 1.07%, and definition increases by 18.8%).
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Wang_C/0/1/0/all/0/1&quot;&gt;Chongyang Wang&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Peng_M/0/1/0/all/0/1&quot;&gt;Ming Peng&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Xu_L/0/1/0/all/0/1&quot;&gt;Lingfeng Xu&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Chen_T/0/1/0/all/0/1&quot;&gt;Tong Chen&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08163">
<title>Multiple target tracking based on sets of trajectories. (arXiv:1605.08163v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08163</link>
<description rdf:parseType="Literal">&lt;p&gt;This paper proposes the set of target trajectories as the state variable for
multiple target tracking. The main objective of multiple target tracking is to
estimate an unknown number of target trajectories given a sequence of
measurements. This quantity of interest is perfectly represented as a set of
trajectories without the need of arbitrary parameters such as labels or
ordering. We use finite-set statistics to pose this problem in the Bayesian
framework and formulate a state space model where the random state is a random
finite set that contains trajectories. All information of interest is thus
contained in the multitrajectory filtering probability density function (PDF),
which represents the multitrajectory PDF of the set of trajectories given the
measurements. For the standard measurement and dynamic models, we describe a
family of PDFs that is conjugate in the sense that the multitrajectory
filtering PDF remains within that family during both the prediction and update
steps.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Garcia_Fernandez_A/0/1/0/all/0/1&quot;&gt;&amp;#xc1;ngel F. Garc&amp;#xed;a-Fern&amp;#xe1;ndez&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Svensson_L/0/1/0/all/0/1&quot;&gt;Lennart Svensson&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Morelande_M/0/1/0/all/0/1&quot;&gt;Mark R. Morelande&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08179">
<title>Discovering Causal Signals in Images. (arXiv:1605.08179v1 [stat.ML])</title>
<link>http://arxiv.org/abs/1605.08179</link>
<description rdf:parseType="Literal">&lt;p&gt;The purpose of this paper is to point out and assay observable causal signals
within collections of static images. We achieve this goal in two steps. First,
we take a learning approach to observational causal inference, and build a
classifier that achieves state-of-the-art performance on finding the causal
direction between pairs of random variables, when given samples from their
joint distribution. Second, we use our causal direction finder to effectively
distinguish between features of objects and features of their contexts in
collections of static images. Our experiments demonstrate the existence of (1)
a relation between the direction of causality and the difference between
objects and their contexts, and (2) observable causal signals in collections of
static images.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/stat/1/au:+Lopez_Paz_D/0/1/0/all/0/1&quot;&gt;David Lopez-Paz&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/stat/1/au:+Nishihara_R/0/1/0/all/0/1&quot;&gt;Robert Nishihara&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/stat/1/au:+Chintala_S/0/1/0/all/0/1&quot;&gt;Soumith Chintala&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/stat/1/au:+Scholkopf_B/0/1/0/all/0/1&quot;&gt;Bernhard Sch&amp;#xf6;lkopf&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/stat/1/au:+Bottou_L/0/1/0/all/0/1&quot;&gt;L&amp;#xe9;on Bottou&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08247">
<title>cvpaper.challenge in 2015 - A review of CVPR2015 and DeepSurvey. (arXiv:1605.08247v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08247</link>
<description rdf:parseType="Literal">&lt;p&gt;The &quot;cvpaper.challenge&quot; is a group composed of members from AIST, Tokyo Denki
Univ. (TDU), and Univ. of Tsukuba that aims to systematically summarize papers
on computer vision, pattern recognition, and related fields. For this
particular review, we focused on reading the ALL 602 conference papers
presented at the CVPR2015, the premier annual computer vision event held in
June 2015, in order to grasp the trends in the field. Further, we are proposing
&quot;DeepSurvey&quot; as a mechanism embodying the entire process from the reading
through all the papers, the generation of ideas, and to the writing of paper.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Kataoka_H/0/1/0/all/0/1&quot;&gt;Hirokatsu Kataoka&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Miyashita_Y/0/1/0/all/0/1&quot;&gt;Yudai Miyashita&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Yamabe_T/0/1/0/all/0/1&quot;&gt;Tomoaki Yamabe&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Shirakabe_S/0/1/0/all/0/1&quot;&gt;Soma Shirakabe&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Sato_S/0/1/0/all/0/1&quot;&gt;Shin&amp;#x27;ichi Sato&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Hoshino_H/0/1/0/all/0/1&quot;&gt;Hironori Hoshino&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Kato_R/0/1/0/all/0/1&quot;&gt;Ryo Kato&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Abe_K/0/1/0/all/0/1&quot;&gt;Kaori Abe&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Imanari_T/0/1/0/all/0/1&quot;&gt;Takaaki Imanari&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Kobayashi_N/0/1/0/all/0/1&quot;&gt;Naomichi Kobayashi&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Morita_S/0/1/0/all/0/1&quot;&gt;Shinichiro Morita&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Nakamura_A/0/1/0/all/0/1&quot;&gt;Akio Nakamura&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08283">
<title>Discrete Deep Feature Extraction: A Theory and New Architectures. (arXiv:1605.08283v1 [cs.LG])</title>
<link>http://arxiv.org/abs/1605.08283</link>
<description rdf:parseType="Literal">&lt;p&gt;First steps towards a mathematical theory of deep convolutional neural
networks for feature extraction were made---for the continuous-time case---in
Mallat, 2012, and Wiatowski and B\&quot;olcskei, 2015. This paper considers the
discrete case, introduces new convolutional neural network architectures, and
proposes a mathematical framework for their analysis. Specifically, we
establish deformation and translation sensitivity results of local and global
nature, and we investigate how certain structural properties of the input
signal are reflected in the corresponding feature vectors. Our theory applies
to general filters and general Lipschitz-continuous non-linearities and pooling
operators. Experiments on handwritten digit classification and facial landmark
detection---including feature importance evaluation---complement the
theoretical findings.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Wiatowski_T/0/1/0/all/0/1&quot;&gt;Thomas Wiatowski&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Tschannen_M/0/1/0/all/0/1&quot;&gt;Michael Tschannen&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Stanic_A/0/1/0/all/0/1&quot;&gt;Aleksandar Stani&amp;#x107;&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Grohs_P/0/1/0/all/0/1&quot;&gt;Philipp Grohs&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Bolcskei_H/0/1/0/all/0/1&quot;&gt;Helmut B&amp;#xf6;lcskei&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08313">
<title>A Light-powered, Always-On, Smart Camera with Compressed Domain Gesture Detection. (arXiv:1605.08313v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08313</link>
<description rdf:parseType="Literal">&lt;p&gt;In this paper we propose an energy-e?cient camera-based gesture recognition
system powered by light energy for \al- ways on&quot; applications. Low energy
consumption is achieved by directly extracting gesture features from the
compressed measurements, which are the block averages and the linear
combinations of the image sensor&apos;s pixel values. The ges- tures are recognized
using a nearest-neighbour (NN) classi- ?er followed by Dynamic Time Warping
(DTW). The sys- tem has been implemented on an Analog Devices Black Fin ULP
vision processor and powered by PV cells whose output is regulated by TI&apos;s
DC-DC buck converter with Maximum Power Point Tracking (MPPT). Measured data
reveals that with only 400 compressed measurements (768? compression ratio) per
frame, the system is able to recognize key wake- up gestures with greater than
80% accuracy and only 95mJ of energy per frame. Owing to its fully self-powered
op- eration, the proposed system can ?nd wide applications in \always-on&quot;
vision systems such as in surveillance, robotics and consumer electronics with
touch-less operation.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+A_A/0/1/0/all/0/1&quot;&gt;Anvesha A&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Xu_S/0/1/0/all/0/1&quot;&gt;Shaojie Xu&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Cao_N/0/1/0/all/0/1&quot;&gt;Ningyuan Cao&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Romberg_J/0/1/0/all/0/1&quot;&gt;Justin Romberg&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Raychowdhury_A/0/1/0/all/0/1&quot;&gt;Arijit Raychowdhury&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08323">
<title>Aerial image geolocalization from recognition and matching of roads and intersections. (arXiv:1605.08323v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08323</link>
<description rdf:parseType="Literal">&lt;p&gt;Aerial image analysis at a semantic level is important in many applications
with strong potential impact in industry and consumer use, such as automated
mapping, urban planning, real estate and environment monitoring, or disaster
relief. The problem is enjoying a great interest in computer vision and remote
sensing, due to increased computer power and improvement in automated image
understanding algorithms. In this paper we address the task of automatic
geolocalization of aerial images from recognition and matching of roads and
intersections. Our proposed method is a novel contribution in the literature
that could enable many applications of aerial image analysis when GPS data is
not available. We offer a complete pipeline for geolocalization, from the
detection of roads and intersections, to the identification of the enclosing
geographic region by matching detected intersections to previously learned
manually labeled ones, followed by accurate geometric alignment between the
detected roads and the manually labeled maps. We test on a novel dataset with
aerial images of two European cities and use the publicly available
OpenStreetMap project for collecting ground truth roads annotations. We show in
extensive experiments that our approach produces highly accurate localizations
in the challenging case when we train on images from one city and test on the
other and the quality of the aerial images is relatively poor. We also show
that the the alignment between detected roads and pre-stored manual annotations
can be effectively used for improving the quality of the road detection
results.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Costea_D/0/1/0/all/0/1&quot;&gt;Dragos Costea&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Leordeanu_M/0/1/0/all/0/1&quot;&gt;Marius Leordeanu&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08350">
<title>Benign-Malignant Lung Nodule Classification with Geometric and Appearance Histogram Features. (arXiv:1605.08350v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08350</link>
<description rdf:parseType="Literal">&lt;p&gt;Lung cancer accounts for the highest number of cancer deaths globally. Early
diagnosis of lung nodules is very important to reduce the mortality rate of
patients by improving the diagnosis and treatment of lung cancer. This work
proposes an automated system to classify lung nodules as malignant and benign
in CT images. It presents extensive experimental results using a combination of
geometric and histogram lung nodule image features and different linear and
non-linear discriminant classifiers. The proposed approach is experimentally
validated on the LIDC-IDRI public lung cancer screening thoracic computed
tomography (CT) dataset containing nodule level diagnostic data. The obtained
results are very encouraging correctly classifying 82% of malignant and 93% of
benign nodules on unseen test data at best.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Shewaye_T/0/1/0/all/0/1&quot;&gt;Tizita Nesibu Shewaye&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Mekonnen_A/0/1/0/all/0/1&quot;&gt;Alhayat Ali Mekonnen&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08359">
<title>Pairwise Decomposition of Image Sequences for Active Multi-View Recognition. (arXiv:1605.08359v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08359</link>
<description rdf:parseType="Literal">&lt;p&gt;A multi-view image sequence provides a much richer capacity for object
recognition than from a single image. However, most existing solutions to
multi-view recognition typically adopt hand-crafted, model-based geometric
methods, which do not readily embrace recent trends in deep learning. We
propose to bring Convolutional Neural Networks to generic multi-view
recognition, by decomposing an image sequence into a set of image pairs,
classifying each pair independently, and then learning an object classifier by
weighting the contribution of each pair. This allows for recognition over
arbitrary camera trajectories, without requiring explicit training over the
potentially infinite number of camera paths and lengths. Building these
pairwise relationships then naturally extends to the next-best-view problem in
an active recognition framework. To achieve this, we train a second
Convolutional Neural Network to map directly from an observed image to next
viewpoint. Finally, we incorporate this into a trajectory optimisation task,
whereby the best recognition confidence is sought for a given trajectory
length. We present state-of-the-art results in both guided and unguided
multi-view recognition on the ModelNet dataset, and show how our method can be
used with depth images, greyscale images, or both.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Johns_E/0/1/0/all/0/1&quot;&gt;Edward Johns&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Leutenegger_S/0/1/0/all/0/1&quot;&gt;Stefan Leutenegger&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Davison_A/0/1/0/all/0/1&quot;&gt;Andrew J. Davison&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08397">
<title>Domain Transfer Multi-Instance Dictionary Learning. (arXiv:1605.08397v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08397</link>
<description rdf:parseType="Literal">&lt;p&gt;In this paper, we invest the domain transfer learning problem with
multi-instance data. We assume we already have a well-trained multi-instance
dictionary and its corresponding classifier from the source domain, which can
be used to represent and classify the bags. But it cannot be directly used to
the target domain. Thus we propose to adapt them to the target domain by adding
an adaptive term to the source domain classifier. The adaptive function is a
linear function based a domain transfer multi-instance dictionary. Given a
target domain bag, we first map it to a bag-level feature space using the
domain transfer dictionary, and then apply a the linear adaptive function to
its bag-level feature vector. To learn the domain-transfer dictionary and the
adaptive function parameter, we simultaneously minimize the average
classification error of the target domain classifier over the target domain
training set, and the complexities of both the adaptive function parameter and
the domain transfer dictionary. The minimization problem is solved by an
iterative algorithm which update the dictionary and the function parameter
alternately. Experiments over several benchmark data sets show the advantage of
the proposed method over existing state-of-the-art domain transfer
multi-instance learning methods.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Wang_K/0/1/0/all/0/1&quot;&gt;Ke Wang&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Liu_J/0/1/0/all/0/1&quot;&gt;Jiayong Liu&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Gonzalez_D/0/1/0/all/0/1&quot;&gt;Daniel Gonz&amp;#xe1;lez&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08401">
<title>Dense Volume-to-Volume Vascular Boundary Detection. (arXiv:1605.08401v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08401</link>
<description rdf:parseType="Literal">&lt;p&gt;In this work, we present a novel 3D-Convolutional Neural Network (CNN)
architecture called I2I-3D that predicts boundary location in volumetric data.
Our fine-to-fine, deeply supervised framework addresses three critical issues
to 3D boundary detection: (1) efficient, holistic, end-to-end volumetric label
training and prediction (2) precise voxel-level prediction to capture fine
scale structures prevalent in medical data and (3) directed multi-scale,
multi-level feature learning. We evaluate our approach on a dataset consisting
of 93 medical image volumes with a wide variety of anatomical regions and
vascular structures. In the process, we also introduce HED-3D, a 3D extension
of the state-of-the-art 2D edge detector (HED). We show that our deep learning
approach out-performs, the current state-of-the-art in 3D vascular boundary
detection (structured forests 3D), by a large margin, as well as HED applied to
slices, and HED-3D while successfully localizing fine structures. With our
approach, boundary detection takes about one minute on a typical 512x512x512
volume.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Merkow_J/0/1/0/all/0/1&quot;&gt;Jameson Merkow&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Kriegman_D/0/1/0/all/0/1&quot;&gt;David Kriegman&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Marsden_A/0/1/0/all/0/1&quot;&gt;Alison Marsden&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Tu_Z/0/1/0/all/0/1&quot;&gt;Zhuowen Tu&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.08412">
<title>CITlab ARGUS for historical handwritten documents. (arXiv:1605.08412v1 [cs.CV])</title>
<link>http://arxiv.org/abs/1605.08412</link>
<description rdf:parseType="Literal">&lt;p&gt;We describe CITlab&apos;s recognition system for the HTRtS competition attached to
the 13. International Conference on Document Analysis and Recognition, ICDAR
2015. The task comprises the recognition of historical handwritten documents.
The core algorithms of our system are based on multi-dimensional recurrent
neural networks (MDRNN) and connectionist temporal classification (CTC). The
software modules behind that as well as the basic utility technologies are
essentially powered by PLANET&apos;s ARGUS framework for intelligent text
recognition and image processing.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Leifert_G/0/1/0/all/0/1&quot;&gt;Gundram Leifert&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Strauss_T/0/1/0/all/0/1&quot;&gt;Tobias Strau&amp;#xdf;&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Gruning_T/0/1/0/all/0/1&quot;&gt;Tobias Gr&amp;#xfc;ning&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Labahn_R/0/1/0/all/0/1&quot;&gt;Roger Labahn&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1602.02660">
<title>Exploiting Cyclic Symmetry in Convolutional Neural Networks. (arXiv:1602.02660v2 [cs.LG] UPDATED)</title>
<link>http://arxiv.org/abs/1602.02660</link>
<description rdf:parseType="Literal">&lt;p&gt;Many classes of images exhibit rotational symmetry. Convolutional neural
networks are sometimes trained using data augmentation to exploit this, but
they are still required to learn the rotation equivariance properties from the
data. Encoding these properties into the network architecture, as we are
already used to doing for translation equivariance by using convolutional
layers, could result in a more efficient use of the parameter budget by
relieving the model from learning them. We introduce four operations which can
be inserted into neural network models as layers, and which can be combined to
make these models partially equivariant to rotations. They also enable
parameter sharing across different orientations. We evaluate the effect of
these architectural modifications on three datasets which exhibit rotational
symmetry and demonstrate improved performance with smaller models.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Dieleman_S/0/1/0/all/0/1&quot;&gt;Sander Dieleman&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Fauw_J/0/1/0/all/0/1&quot;&gt;Jeffrey De Fauw&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Kavukcuoglu_K/0/1/0/all/0/1&quot;&gt;Koray Kavukcuoglu&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1602.02720">
<title>Multimodal Remote Sensing Image Registration with Accuracy Estimation at Local and Global Scales. (arXiv:1602.02720v2 [cs.CV] UPDATED)</title>
<link>http://arxiv.org/abs/1602.02720</link>
<description rdf:parseType="Literal">&lt;p&gt;This paper focuses on potential accuracy of remote sensing images
registration. We investigate how this accuracy can be estimated without ground
truth available and used to improve registration quality of mono- and
multi-modal pair of images. At the local scale of image fragments, the
Cramer-Rao lower bound (CRLB) on registration error is estimated for each local
correspondence between coarsely registered pair of images. This CRLB is defined
by local image texture and noise properties. Opposite to the standard approach,
where registration accuracy is only evaluated at the output of the registration
process, such valuable information is used by us as an additional input
knowledge. It greatly helps detecting and discarding outliers and refining the
estimation of geometrical transformation model parameters. Based on these
ideas, a new area-based registration method called RAE (Registration with
Accuracy Estimation) is proposed. In addition to its ability to automatically
register very complex multimodal image pairs with high accuracy, the RAE method
provides registration accuracy at the global scale as covariance matrix of
estimation error of geometrical transformation model parameters or as
point-wise registration Standard Deviation. This accuracy does not depend on
any ground truth availability and characterizes each pair of registered images
individually. Thus, the RAE method can identify image areas for which a
predefined registration accuracy is guaranteed. The RAE method is proved
successful with reaching subpixel accuracy while registering eight complex
mono/multimodal and multitemporal image pairs including optical to optical,
optical to radar, optical to Digital Elevation Model (DEM) images and DEM to
radar cases. Other methods employed in comparisons fail to provide in a stable
manner accurate results on the same test cases.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Uss_M/0/1/0/all/0/1&quot;&gt;M.L. Uss&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Vozel_B/0/1/0/all/0/1&quot;&gt;B. Vozel&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Lukin_V/0/1/0/all/0/1&quot;&gt;V.V. Lukin&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Chehdi_K/0/1/0/all/0/1&quot;&gt;K. Chehdi&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1602.05531">
<title>On the Use of Deep Learning for Blind Image Quality Assessment. (arXiv:1602.05531v3 [cs.CV] UPDATED)</title>
<link>http://arxiv.org/abs/1602.05531</link>
<description rdf:parseType="Literal">&lt;p&gt;In this work we investigate the use of deep learning for distortion-generic
blind image quality assessment. We report on different design choices, ranging
from the use of features extracted from pre-trained Convolutional Neural
Networks (CNNs) as a generic image description, to the use of features
extracted from a CNN fine-tuned for the image quality task. Our best proposal,
named DeepBIQ, estimates the image quality by average pooling the scores
predicted on multiple sub-regions of the original image. The score of each
sub-region is computed using a Support Vector Regression (SVR) machine taking
as input features extracted using a CNN fine-tuned for category-based image
quality assessment. Experimental results on the LIVE In the Wild Image Quality
Challenge Database and on the LIVE Image Quality Assessment Database show that
DeepBIQ outperforms the state-of-the-art methods compared, having a Linear
Correlation Coefficient (LCC) with human subjective scores of almost 0.91 and
0.98 respectively. Furthermore, in most of the cases, the quality score
predictions of DeepBIQ are closer to the average observer than those of a
generic human observer.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Bianco_S/0/1/0/all/0/1&quot;&gt;Simone Bianco&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Celona_L/0/1/0/all/0/1&quot;&gt;Luigi Celona&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Napoletano_P/0/1/0/all/0/1&quot;&gt;Paolo Napoletano&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Schettini_R/0/1/0/all/0/1&quot;&gt;Raimondo Schettini&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1604.07666">
<title>$\ell_p$-Box ADMM: A Versatile Framework for Integer Programming. (arXiv:1604.07666v2 [cs.CV] UPDATED)</title>
<link>http://arxiv.org/abs/1604.07666</link>
<description rdf:parseType="Literal">&lt;p&gt;This paper revisits the integer programming (IP) problem, which plays a
fundamental role in many computer vision and machine learning applications. The
literature abounds with many seminal works that address this problem, some
focusing on continuous approaches (e.g. linear program relaxation) while others
on discrete ones (e.g., min-cut). However, a limited number of them are
designed to handle the general IP form and even these methods cannot adequately
satisfy the simultaneous requirements of accuracy, feasibility, and
scalability. To this end, we propose a novel and versatile framework called
$\ell_p$-box ADMM, which is based on two parts. (1) The discrete constraint is
equivalently replaced by the intersection of a box and a $(n-1)$-dimensional
sphere (defined through the $\ell_p$ norm). (2) We infuse this equivalence into
the ADMM (Alternating Direction Method of Multipliers) framework to handle
these continuous constraints separately and to harness its attractive
properties. More importantly, the ADMM update steps can lead to manageable
sub-problems in the continuous domain. To demonstrate its efficacy, we consider
an instance of the framework, namely $\ell_2$-box ADMM applied to binary
quadratic programming (BQP). Here, the ADMM steps are simple, computationally
efficient, and theoretically guaranteed to converge to a KKT point. We
demonstrate the applicability of $\ell_2$-box ADMM on three important
applications: MRF energy minimization, graph matching, and clustering. Results
clearly show that it significantly outperforms existing generic IP solvers both
in runtime and objective. It also achieves very competitive performance vs.
state-of-the-art methods specific to these applications.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Wu_B/0/1/0/all/0/1&quot;&gt;Baoyuan Wu&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Ghanem_B/0/1/0/all/0/1&quot;&gt;Bernard Ghanem&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1604.08806">
<title>3D Interest Point Detection Based on Geometric Measures and Sparse Refinement. (arXiv:1604.08806v2 [cs.CV] UPDATED)</title>
<link>http://arxiv.org/abs/1604.08806</link>
<description rdf:parseType="Literal">&lt;p&gt;Three dimensional (3D) interest point detection plays a fundamental role in
computer vision. In this paper, we introduce a new method for detecting 3D
interest points of 3D mesh models based on geometric measures and sparse
refinement (GMSR). The key point of our approach is to calculate the 3D
saliency measure using two novel geometric measures, which are defined in
multi-scale space to effectively distinguish 3D interest points from edges and
flat areas. Those points with local maxima of 3D saliency measure are selected
as the candidates of 3D interest points. Finally, we utilize an $l_0$ norm
based optimization method to refine the candidates of 3D interest points by
constraining the number of 3D interest points. Numerical experiments show that
the proposed GMSR based 3D interest point detector outperforms current six
state-of-the-art methods for different kinds of 3D mesh models.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Lin_X/0/1/0/all/0/1&quot;&gt;Xinyu Lin&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Zhu_C/0/1/0/all/0/1&quot;&gt;Ce Zhu&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Zhang_Q/0/1/0/all/0/1&quot;&gt;Qian Zhang&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Liu_Y/0/1/0/all/0/1&quot;&gt;Yipeng Liu&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.05579">
<title>Low-Rank Matrices on Graphs: Generalized Recovery &amp; Applications. (arXiv:1605.05579v3 [cs.CV] UPDATED)</title>
<link>http://arxiv.org/abs/1605.05579</link>
<description rdf:parseType="Literal">&lt;p&gt;Many real world datasets subsume a linear or non-linear low-rank structure in
a very low-dimensional space. Unfortunately, one often has very little or no
information about the geometry of the space, resulting in a highly
under-determined recovery problem. Under certain circumstances,
state-of-the-art algorithms provide an exact recovery for linear low-rank
structures but at the expense of highly inscalable algorithms which use nuclear
norm. However, the case of non-linear structures remains unresolved. We revisit
the problem of low-rank recovery from a totally different perspective,
involving graphs which encode pairwise similarity between the data samples and
features. Surprisingly, our analysis confirms that it is possible to recover
many approximate linear and non-linear low-rank structures with recovery
guarantees with a set of highly scalable and efficient algorithms. We call such
data matrices as \textit{Low-Rank matrices on graphs} and show that many real
world datasets satisfy this assumption approximately due to underlying
stationarity. Our detailed theoretical and experimental analysis unveils the
power of the simple, yet very novel recovery framework \textit{Fast Robust PCA
on Graphs}
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Shahid_N/0/1/0/all/0/1&quot;&gt;Nauman Shahid&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Perraudin_N/0/1/0/all/0/1&quot;&gt;Nathanael Perraudin&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Vandergheynst_P/0/1/0/all/0/1&quot;&gt;Pierre Vandergheynst&lt;/a&gt;</dc:creator>
</item>
<item rdf:about="http://arxiv.org/abs/1605.07912">
<title>Encode, Review, and Decode: Reviewer Module for Caption Generation. (arXiv:1605.07912v2 [cs.LG] UPDATED)</title>
<link>http://arxiv.org/abs/1605.07912</link>
<description rdf:parseType="Literal">&lt;p&gt;We propose a novel module, the reviewer module, to improve the
encoder-decoder learning framework. The reviewer module is generic, and can be
plugged into an existing encoder-decoder model. The reviewer module performs a
number of review steps with attention mechanism on the encoder hidden states,
and outputs a fact vector after each review step; the fact vectors are used as
the input of the attention mechanism in the decoder. We show that the
conventional encoder-decoders are a special case of our framework. Empirically,
we show that our framework can improve over state-of-the-art encoder-decoder
systems on the tasks of image captioning and source code captioning.
&lt;/p&gt;
</description>
<dc:creator> &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Yang_Z/0/1/0/all/0/1&quot;&gt;Zhilin Yang&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Yuan_Y/0/1/0/all/0/1&quot;&gt;Ye Yuan&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Wu_Y/0/1/0/all/0/1&quot;&gt;Yuexin Wu&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Salakhutdinov_R/0/1/0/all/0/1&quot;&gt;Ruslan Salakhutdinov&lt;/a&gt;, &lt;a href=&quot;http://arxiv.org/find/cs/1/au:+Cohen_W/0/1/0/all/0/1&quot;&gt;William W. Cohen&lt;/a&gt;</dc:creator>
</item>
</rdf:RDF>