Unnamed: 0.1
int64
0
41k
Unnamed: 0
int64
0
41k
author
stringlengths
9
1.39k
id
stringlengths
11
18
summary
stringlengths
25
3.66k
title
stringlengths
4
258
year
int64
1.99k
2.02k
arxiv_url
stringlengths
32
39
info
stringlengths
523
3.18k
embeddings
stringlengths
16.9k
17.1k
1,900
1,900
['Junhua Mao', 'Wei Xu', 'Yi Yang', 'Jiang Wang', 'Zhiheng Huang', 'Alan Yuille']
1412.6632v5
In this paper, we present a multimodal Recurrent Neural Network (m-RNN) model for generating novel image captions. It directly models the probability distribution of generating a word given previous words and an image. Image captions are generated by sampling from this distribution. The model consists of two sub-networ...
Deep Captioning with Multimodal Recurrent Neural Networks (m-RNN)
2,014
http://arxiv.org/pdf/1412.6632v5
Title Deep Captioning Multimodal Recurrent Neural Networks mRNN Summary paper present multimodal Recurrent Neural Network mRNN model generating novel image caption directly model probability distribution generating word given previous word image Image caption generated sampling distribution model consists two subnetwor...
[0.0793909877538681, 0.04649634659290314, -0.0013939253985881805, 0.04682249575853348, -0.03486587852239609, -0.010868647135794163, 0.010027806274592876, -0.005283457692712545, -0.023180022835731506, -0.05472239479422569, -0.011653653346002102, -0.025220155715942383, -0.008407999761402607, 0.056870460510253906, 0.01624...
1,901
1,901
['Angeliki Lazaridou', 'Nghia The Pham', 'Marco Baroni']
1501.02598v3
We extend the SKIP-GRAM model of Mikolov et al. (2013a) by taking visual information into account. Like SKIP-GRAM, our multimodal models (MMSKIP-GRAM) build vector-based word representations by learning to predict linguistic contexts in text corpora. However, for a restricted set of words, the models are also exposed t...
Combining Language and Vision with a Multimodal Skip-gram Model
2,015
http://arxiv.org/pdf/1501.02598v3
Title Combining Language Vision Multimodal Skipgram Model Summary extend SKIPGRAM model Mikolov et al 2013a taking visual information account Like SKIPGRAM multimodal model MMSKIPGRAM build vectorbased word representation learning predict linguistic context text corpus However restricted set word model also exposed vis...
[0.04908914119005203, 0.01721227541565895, -0.00902736559510231, 0.04834049940109253, -0.02523280307650566, 0.009042426943778992, 0.003951284568756819, 0.041498612612485886, -0.0044868928380310535, -0.07964462041854858, -0.0019926726818084717, -0.01477618794888258, 0.03014116734266281, 0.0454435870051384, 0.05133299902...
1,902
1,902
['Junhua Mao', 'Wei Xu', 'Yi Yang', 'Jiang Wang', 'Zhiheng Huang', 'Alan Yuille']
1504.06692v2
In this paper, we address the task of learning novel visual concepts, and their interactions with other concepts, from a few images with sentence descriptions. Using linguistic context and visual features, our method is able to efficiently hypothesize the semantic meaning of new words and add them to its word dictionar...
Learning like a Child: Fast Novel Visual Concept Learning from Sentence Descriptions of Images
2,015
http://arxiv.org/pdf/1504.06692v2
Title Learning like Child Fast Novel Visual Concept Learning Sentence Descriptions Images Summary paper address task learning novel visual concept interaction concept image sentence description Using linguistic context visual feature method able efficiently hypothesize semantic meaning new word add word dictionary used...
[0.05880361795425415, 0.012007403187453747, -0.004071745090186596, 0.05339255556464195, -0.028941739350557327, 0.03297742083668709, 0.006328505929559469, -0.022559313103556633, -0.03923584520816803, -0.05413109064102173, -0.009392209351062775, -0.001380954752676189, -0.002745207166299224, 0.05428895726799965, 0.0066782...
1,903
1,903
['Samy Bengio', 'Oriol Vinyals', 'Navdeep Jaitly', 'Noam Shazeer']
1506.03099v3
Recurrent Neural Networks can be trained to produce sequences of tokens given some input, as exemplified by recent results in machine translation and image captioning. The current approach to training them consists of maximizing the likelihood of each token in the sequence given the current (recurrent) state and the pr...
Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
2,015
http://arxiv.org/pdf/1506.03099v3
Title Scheduled Sampling Sequence Prediction Recurrent Neural Networks Summary Recurrent Neural Networks trained produce sequence token given input exemplified recent result machine translation image captioning current approach training consists maximizing likelihood token sequence given current recurrent state previou...
[0.04436810314655304, 0.022265905514359474, -0.010869869962334633, 0.017911946401000023, -0.014149378053843975, -0.017918044701218605, 0.01644875481724739, -0.0131453275680542, -0.024018095806241035, -0.03973142057657242, 0.007947895675897598, -0.0200678501278162, 0.013485431671142578, 0.06214265152812004, 0.0204907283...
1,904
1,904
['Junhua Mao', 'Jonathan Huang', 'Alexander Toshev', 'Oana Camburu', 'Alan Yuille', 'Kevin Murphy']
1511.02283v3
We propose a method that can generate an unambiguous description (known as a referring expression) of a specific object or region in an image, and which can also comprehend or interpret such an expression to infer which object is being described. We show that our method outperforms previous methods that generate descri...
Generation and Comprehension of Unambiguous Object Descriptions
2,015
http://arxiv.org/pdf/1511.02283v3
Title Generation Comprehension Unambiguous Object Descriptions Summary propose method generate unambiguous description known referring expression specific object region image also comprehend interpret expression infer object described show method outperforms previous method generate description object without taking ac...
[0.05775875225663185, 0.05294516682624817, 0.018609188497066498, 0.06581497937440872, -0.021189767867326736, 0.005093289073556662, 0.022057348862290382, -0.0020759650506079197, -0.04518275335431099, -0.012719531543552876, 0.002339413622394204, 0.0415329784154892, 0.034309521317481995, 0.033394888043403625, 0.0027420429...
1,905
1,905
['Anna Rohrbach', 'Marcus Rohrbach', 'Ronghang Hu', 'Trevor Darrell', 'Bernt Schiele']
1511.03745v4
Grounding (i.e. localizing) arbitrary, free-form textual phrases in visual content is a challenging problem with many applications for human-computer interaction and image-text reference resolution. Few datasets provide the ground truth spatial localization of phrases, thus it is desirable to learn from data with no or...
Grounding of Textual Phrases in Images by Reconstruction
2,015
http://arxiv.org/pdf/1511.03745v4
Title Grounding Textual Phrases Images Reconstruction Summary Grounding ie localizing arbitrary freeform textual phrase visual content challenging problem many application humancomputer interaction imagetext reference resolution datasets provide ground truth spatial localization phrase thus desirable learn data little ...
[0.05351989343762398, 0.04404808580875397, 0.03433031961321831, 0.04480961710214615, -0.011151671409606934, -0.012128446251153946, 0.0031569357961416245, 0.0028138908091932535, -0.018939824774861336, -0.056779421865940094, -0.012786462903022766, 0.01619798131287098, 0.04010212793946266, 0.011041603982448578, 0.04487297...
1,906
1,906
['Mohamed Elhoseiny', 'Scott Cohen', 'Walter Chang', 'Brian Price', 'Ahmed Elgammal']
1511.04891v4
We study scalable and uniform understanding of facts in images. Existing visual recognition systems are typically modeled differently for each fact type such as objects, actions, and interactions. We propose a setting where all these facts can be modeled simultaneously with a capacity to understand unbounded number of ...
Sherlock: Scalable Fact Learning in Images
2,015
http://arxiv.org/pdf/1511.04891v4
Title Sherlock Scalable Fact Learning Images Summary study scalable uniform understanding fact image Existing visual recognition system typically modeled differently fact type object action interaction propose setting fact modeled simultaneously capacity understand unbounded number fact structured way training data com...
[-0.012998024001717567, -0.014170507900416851, 0.011001886799931526, 0.08346516638994217, -0.0033055003732442856, 0.03764963150024414, 0.021781960502266884, 0.015552295371890068, 0.002036727499216795, -0.04522998258471489, 0.009102778509259224, -0.001654736464843154, -0.027632640674710274, 0.07406899333000183, -0.01063...
1,907
1,907
['Peng Zhang', 'Yash Goyal', 'Douglas Summers-Stay', 'Dhruv Batra', 'Devi Parikh']
1511.05099v5
The complex compositional structure of language makes problems at the intersection of vision and language challenging. But language also provides a strong prior that can result in good superficial performance, without the underlying models truly understanding the visual content. This can hinder progress in pushing stat...
Yin and Yang: Balancing and Answering Binary Visual Questions
2,015
http://arxiv.org/pdf/1511.05099v5
Title Yin Yang Balancing Answering Binary Visual Questions Summary complex compositional structure language make problem intersection vision language challenging language also provides strong prior result good superficial performance without underlying model truly understanding visual content hinder progress pushing st...
[0.04937983676791191, 0.044918257743120193, -0.027279075235128403, 0.04331352934241295, -0.040603213012218475, 0.005199109204113483, 0.002626160392537713, 0.017412172630429268, -0.011667820625007153, -0.054582539945840836, -0.0035920487716794014, 0.0022545072715729475, 0.010929903015494347, 0.11995977908372879, -0.0081...
1,908
1,908
['Hyeonwoo Noh', 'Paul Hongsuck Seo', 'Bohyung Han']
1511.05756v1
We tackle image question answering (ImageQA) problem by learning a convolutional neural network (CNN) with a dynamic parameter layer whose weights are determined adaptively based on questions. For the adaptive parameter prediction, we employ a separate parameter prediction network, which consists of gated recurrent uni...
Image Question Answering using Convolutional Neural Network with Dynamic Parameter Prediction
2,015
http://arxiv.org/pdf/1511.05756v1
Title Image Question Answering using Convolutional Neural Network Dynamic Parameter Prediction Summary tackle image question answering ImageQA problem learning convolutional neural network CNN dynamic parameter layer whose weight determined adaptively based question adaptive parameter prediction employ separate paramet...
[0.051230691373348236, 0.013715321198105812, -0.013175424188375473, 0.056065913289785385, 0.021962478756904602, 0.00792464055120945, 0.01949167624115944, 0.01350838877260685, -0.006171965040266514, -0.010045954957604408, 0.008784470148384571, -0.01721310056746006, -0.03806006908416748, 0.03215565159916878, 0.0292893089...
1,909
1,909
['Liwei Wang', 'Yin Li', 'Svetlana Lazebnik']
1511.06078v2
This paper proposes a method for learning joint embeddings of images and text using a two-branch neural network with multiple layers of linear projections followed by nonlinearities. The network is trained using a large margin objective that combines cross-view ranking constraints with within-view neighborhood structur...
Learning Deep Structure-Preserving Image-Text Embeddings
2,015
http://arxiv.org/pdf/1511.06078v2
Title Learning Deep StructurePreserving ImageText Embeddings Summary paper proposes method learning joint embeddings image text using twobranch neural network multiple layer linear projection followed nonlinearities network trained using large margin objective combine crossview ranking constraint withinview neighborhoo...
[-0.00699905538931489, 0.06008131802082062, 0.019515084102749825, 0.059643302112817764, -0.014183498919010162, 0.011076982133090496, -0.02245064452290535, 0.03937374800443649, 0.03240153193473816, -0.0124816307798028, -0.020078305155038834, -0.02400781214237213, -0.028013093397021294, -0.0005240186583250761, 0.02696070...
1,910
1,910
['Ivan Vendrov', 'Ryan Kiros', 'Sanja Fidler', 'Raquel Urtasun']
1511.06361v6
Hypernymy, textual entailment, and image captioning can be seen as special cases of a single visual-semantic hierarchy over words, sentences, and images. In this paper we advocate for explicitly modeling the partial order structure of this hierarchy. Towards this goal, we introduce a general method for learning ordered...
Order-Embeddings of Images and Language
2,015
http://arxiv.org/pdf/1511.06361v6
Title OrderEmbeddings Images Language Summary Hypernymy textual entailment image captioning seen special case single visualsemantic hierarchy word sentence image paper advocate explicitly modeling partial order structure hierarchy Towards goal introduce general method learning ordered representation show applied variet...
[0.0598696693778038, 0.050149306654930115, -0.02525370568037033, 0.07718096673488617, 0.009937568567693233, 0.033298347145318985, -0.009070387110114098, 0.017352713271975517, 0.011572662740945816, -0.051530979573726654, 0.011028998531401157, -0.02919090911746025, -0.0012201991630718112, 0.009193511679768562, 0.00341075...
1,911
1,911
['Mohamed Elhoseiny', 'Jingen Liu', 'Hui Cheng', 'Harpreet Sawhney', 'Ahmed Elgammal']
1512.00818v2
We propose a new zero-shot Event Detection method by Multi-modal Distributional Semantic embedding of videos. Our model embeds object and action concepts as well as other available modalities from videos into a distributional semantic space. To our knowledge, this is the first Zero-Shot event detection model that is bu...
Zero-Shot Event Detection by Multimodal Distributional Semantic Embedding of Videos
2,015
http://arxiv.org/pdf/1512.00818v2
Title ZeroShot Event Detection Multimodal Distributional Semantic Embedding Videos Summary propose new zeroshot Event Detection method Multimodal Distributional Semantic embedding video model embeds object action concept well available modality video distributional semantic space knowledge first ZeroShot event detectio...
[-0.031106218695640564, 0.016432391479611397, 0.01689792051911354, 0.05312442407011986, -0.011434118263423443, -0.0168935377150774, -0.001090610632672906, 0.05972377583384514, -0.04854621738195419, -0.11797219514846802, 0.03344038873910904, 0.03859531879425049, -0.03667730838060379, 0.06557496637105942, 0.0012652415316...
1,912
1,912
['Arjun Chandrasekaran', 'Ashwin K. Vijayakumar', 'Stanislaw Antol', 'Mohit Bansal', 'Dhruv Batra', 'C. Lawrence Zitnick', 'Devi Parikh']
1512.04407v4
Humor is an integral part of human lives. Despite being tremendously impactful, it is perhaps surprising that we do not have a detailed understanding of humor yet. As interactions between humans and AI systems increase, it is imperative that these systems are taught to understand subtleties of human expressions such as...
We Are Humor Beings: Understanding and Predicting Visual Humor
2,015
http://arxiv.org/pdf/1512.04407v4
Title Humor Beings Understanding Predicting Visual Humor Summary Humor integral part human life Despite tremendously impactful perhaps surprising detailed understanding humor yet interaction human AI system increase imperative system taught understand subtlety human expression humor work interested question content sce...
[0.08771549165248871, 0.07180382311344147, -0.03546302393078804, -0.009996481239795685, 0.01573587954044342, 0.04774585738778114, 0.036728233098983765, 0.04824230819940567, 0.04828201234340668, -0.03717821091413498, 0.009566275402903557, 0.03356766700744629, 2.5227589503629133e-05, 0.04144928976893425, 0.02593978121876...
1,913
1,913
['Mohamed Elhoseiny', 'Ahmed Elgammal', 'Babak Saleh']
1601.00025v2
People typically learn through exposure to visual concepts associated with linguistic descriptions. For instance, teaching visual object categories to children is often accompanied by descriptions in text or speech. In a machine learning context, these observations motivates us to ask whether this learning process coul...
Write a Classifier: Predicting Visual Classifiers from Unstructured Text
2,015
http://arxiv.org/pdf/1601.00025v2
Title Write Classifier Predicting Visual Classifiers Unstructured Text Summary People typically learn exposure visual concept associated linguistic description instance teaching visual object category child often accompanied description text speech machine learning context observation motivates u ask whether learning p...
[0.03654329106211662, 0.018298231065273285, -0.03063666634261608, 0.044718410819768906, -0.009789991192519665, 0.016778774559497833, -0.020594799891114235, 0.038557227700948715, 0.013630416244268417, -0.10864344984292984, 0.026198159903287888, 0.0425034761428833, 0.01608813740313053, 0.043444402515888214, 0.00070932914...
1,914
1,914
['Afroze Ibrahim Baqapuri']
1601.03478v1
The ability to describe images with natural language sentences is the hallmark for image and language understanding. Such a system has wide ranging applications such as annotating images and using natural sentences to search for images.In this project we focus on the task of bidirectional image retrieval: such asystem ...
Deep Learning Applied to Image and Text Matching
2,015
http://arxiv.org/pdf/1601.03478v1
Title Deep Learning Applied Image Text Matching Summary ability describe image natural language sentence hallmark image language understanding system wide ranging application annotating image using natural sentence search imagesIn project focus task bidirectional image retrieval asystem capable retrieving image based s...
[0.06130256503820419, 0.03314635902643204, 0.007418754510581493, 0.10536203533411026, -0.0591852180659771, 0.01437277439981699, 0.027494970709085464, 0.021089883521199226, -0.009528565220534801, -0.038062889128923416, -0.020498190075159073, 0.004370720591396093, -0.037772659212350845, 0.05724097788333893, 0.02649604529...
1,915
1,915
['Shijian Tang', 'Song Han']
1602.01895v1
Generating natural language descriptions for images is a challenging task. The traditional way is to use the convolutional neural network (CNN) to extract image features, followed by recurrent neural network (RNN) to generate sentences. In this paper, we present a new model that added memory cells to gate the feeding o...
Generate Image Descriptions based on Deep RNN and Memory Cells for Images Features
2,016
http://arxiv.org/pdf/1602.01895v1
Title Generate Image Descriptions based Deep RNN Memory Cells Images Features Summary Generating natural language description image challenging task traditional way use convolutional neural network CNN extract image feature followed recurrent neural network RNN generate sentence paper present new model added memory cel...
[0.046134982258081436, 0.0635976567864418, 0.02595936506986618, 0.07111554592847824, -0.04028385132551193, 0.004777907859534025, 0.0203280970454216, -0.035092491656541824, -0.021020811051130295, -0.03534916415810585, -0.01842723973095417, -0.03319365531206131, 0.015593559481203556, 0.061000265181064606, 0.0088247591629...
1,916
1,916
['Linlin Chao', 'Jianhua Tao', 'Minghao Yang', 'Ya Li', 'Zhengqi Wen']
1603.08321v1
This paper focuses on two key problems for audio-visual emotion recognition in the video. One is the audio and visual streams temporal alignment for feature level fusion. The other one is locating and re-weighting the perception attentions in the whole audio-visual stream for better recognition. The Long Short Term Mem...
Audio Visual Emotion Recognition with Temporal Alignment and Perception Attention
2,016
http://arxiv.org/pdf/1603.08321v1
Title Audio Visual Emotion Recognition Temporal Alignment Perception Attention Summary paper focus two key problem audiovisual emotion recognition video One audio visual stream temporal alignment feature level fusion one locating reweighting perception attention whole audiovisual stream better recognition Long Short Te...
[-0.021154507994651794, -0.05716639384627342, 0.016844959929585457, 0.054869573563337326, 0.03188221529126167, -0.020258570089936256, -0.0015378228854387999, -0.050759173929691315, -0.02952943928539753, -0.0239292923361063, -0.060691773891448975, -0.009757154621183872, 0.008805305697023869, 0.027531523257493973, -0.023...
1,917
1,917
['Angeliki Lazaridou', 'Nghia The Pham', 'Marco Baroni']
1605.07133v1
We propose an interactive multimodal framework for language learning. Instead of being passively exposed to large amounts of natural text, our learners (implemented as feed-forward neural networks) engage in cooperative referential games starting from a tabula rasa setup, and thus develop their own language from the ne...
Towards Multi-Agent Communication-Based Language Learning
2,016
http://arxiv.org/pdf/1605.07133v1
Title Towards MultiAgent CommunicationBased Language Learning Summary propose interactive multimodal framework language learning Instead passively exposed large amount natural text learner implemented feedforward neural network engage cooperative referential game starting tabula rasa setup thus develop language need co...
[0.05331356078386307, 0.010658401064574718, -0.03818010166287422, 0.05020883306860924, -0.017353946343064308, 0.003138753119856119, 0.01735346019268036, -0.02655145898461342, 0.022989239543676376, -0.06079627573490143, -0.037776995450258255, 0.007051250897347927, 0.008732425048947334, 0.041848354041576385, 0.0360692292...
1,918
1,918
['Zhilin Yang', 'Ye Yuan', 'Yuexin Wu', 'Ruslan Salakhutdinov', 'William W. Cohen']
1605.07912v4
We propose a novel extension of the encoder-decoder framework, called a review network. The review network is generic and can enhance any existing encoder- decoder model: in this paper, we consider RNN decoders with both CNN and RNN encoders. The review network performs a number of review steps with attention mechanism...
Review Networks for Caption Generation
2,016
http://arxiv.org/pdf/1605.07912v4
Title Review Networks Caption Generation Summary propose novel extension encoderdecoder framework called review network review network generic enhance existing encoder decoder model paper consider RNN decoder CNN RNN encoders review network performs number review step attention mechanism encoder hidden state output tho...
[0.05868959054350853, 0.0289948508143425, 0.0014745807275176048, 0.03004823997616768, 0.008412117138504982, 0.01886371523141861, -0.0008221014868468046, 0.0062565552070736885, -0.05956875905394554, -0.027924291789531708, -0.011637681163847446, 0.018915768712759018, 0.016632039099931717, 0.08525408804416656, -0.00761369...
1,919
1,919
['Nazneen Fatema Rajani', 'Raymond J. Mooney']
1605.08764v1
Ensembling methods are well known for improving prediction accuracy. However, they are limited in the sense that they cannot discriminate among component models effectively. In this paper, we propose stacking with auxiliary features that learns to fuse relevant information from multiple systems to improve performance. ...
Stacking With Auxiliary Features
2,016
http://arxiv.org/pdf/1605.08764v1
Title Stacking Auxiliary Features Summary Ensembling method well known improving prediction accuracy However limited sense cannot discriminate among component model effectively paper propose stacking auxiliary feature learns fuse relevant information multiple system improve performance Auxiliary feature enable stacker ...
[0.020187320187687874, 0.0305443424731493, 0.004901321139186621, 0.06804125756025314, 0.024295510724186897, 0.055930640548467636, 0.020448604598641396, 0.018606876954436302, -0.010276173241436481, -0.0244930237531662, -0.020152101293206215, -0.021450351923704147, 0.03347921743988991, 0.04576176777482033, 0.011288509704...
1,920
1,920
['Chenxi Liu', 'Junhua Mao', 'Fei Sha', 'Alan Yuille']
1605.09553v2
Attention mechanisms have recently been introduced in deep learning for various tasks in natural language processing and computer vision. But despite their popularity, the "correctness" of the implicitly-learned attention maps has only been assessed qualitatively by visualization of several examples. In this paper we f...
Attention Correctness in Neural Image Captioning
2,016
http://arxiv.org/pdf/1605.09553v2
Title Attention Correctness Neural Image Captioning Summary Attention mechanism recently introduced deep learning various task natural language processing computer vision despite popularity correctness implicitlylearned attention map assessed qualitatively visualization several example paper focus evaluating improving ...
[0.07229209691286087, 0.0192995797842741, -0.009313875809311867, 0.027889646589756012, 0.016291910782456398, 0.008356530219316483, 0.012270179577171803, -0.019264928996562958, 0.019167259335517883, -0.05160246789455414, -0.01764633134007454, -0.011989313177764416, 0.020814115181565285, 0.035044509917497635, 0.034597128...
1,921
1,921
['Arijit Ray', 'Gordon Christie', 'Mohit Bansal', 'Dhruv Batra', 'Devi Parikh']
1606.06622v3
Visual Question Answering (VQA) is the task of answering natural-language questions about images. We introduce the novel problem of determining the relevance of questions to images in VQA. Current VQA models do not reason about whether a question is even related to the given image (e.g. What is the capital of Argentina...
Question Relevance in VQA: Identifying Non-Visual And False-Premise Questions
2,016
http://arxiv.org/pdf/1606.06622v3
Title Question Relevance VQA Identifying NonVisual FalsePremise Questions Summary Visual Question Answering VQA task answering naturallanguage question image introduce novel problem determining relevance question image VQA Current VQA model reason whether question even related given image eg capital Argentina requires ...
[0.0675911232829094, 0.006879306863993406, -0.032082945108413696, 0.018711306154727936, -0.03669046238064766, 0.022293323650956154, 0.020649926736950874, 0.028358710929751396, 0.018001053482294083, -0.032998085021972656, -0.0037338114343583584, -0.014373973943293095, -0.008558426052331924, 0.04846183955669403, 0.015763...
1,922
1,922
['Hao Zhang', 'Zhiting Hu', 'Yuntian Deng', 'Mrinmaya Sachan', 'Zhicheng Yan', 'Eric P. Xing']
1606.09239v1
We study the problem of automatically building hypernym taxonomies from textual and visual data. Previous works in taxonomy induction generally ignore the increasingly prominent visual data, which encode important perceptual semantics. Instead, we propose a probabilistic model for taxonomy induction by jointly leveragi...
Learning Concept Taxonomies from Multi-modal Data
2,016
http://arxiv.org/pdf/1606.09239v1
Title Learning Concept Taxonomies Multimodal Data Summary study problem automatically building hypernym taxonomy textual visual data Previous work taxonomy induction generally ignore increasingly prominent visual data encode important perceptual semantics Instead propose probabilistic model taxonomy induction jointly l...
[0.029055966064333916, 0.03925388678908348, -0.007833646610379219, 0.07040741294622421, 0.005650457926094532, 0.0314328595995903, -0.014002575539052486, 0.017771225422620773, -0.038794081658124924, -0.09075689315795898, -0.009151303209364414, -0.0007647841703146696, 0.00769038125872612, 0.061802156269550323, 0.02718330...
1,923
1,923
['Matthias Plappert', 'Christian Mandery', 'Tamim Asfour']
1607.03827v1
Linking human motion and natural language is of great interest for the generation of semantic representations of human activities as well as for the generation of robot activities based on natural language input. However, while there have been years of research in this area, no standardized and openly available dataset...
The KIT Motion-Language Dataset
2,016
http://arxiv.org/pdf/1607.03827v1
Title KIT MotionLanguage Dataset Summary Linking human motion natural language great interest generation semantic representation human activity well generation robot activity based natural language input However year research area standardized openly available dataset exists support development evaluation system theref...
[0.016659513115882874, 0.009078282862901688, 0.003103096503764391, 0.00870544370263815, 0.0012981063919141889, 0.03728282079100609, 0.01853986270725727, 0.00388489686883986, -0.018764259293675423, -0.029823871329426765, -0.012982301414012909, 0.0059290057979524136, 0.04428849741816521, 0.012053892016410828, 0.026149356...
1,924
1,924
['Yannis M. Assael', 'Brendan Shillingford', 'Shimon Whiteson', 'Nando de Freitas']
1611.01599v2
Lipreading is the task of decoding text from the movement of a speaker's mouth. Traditional approaches separated the problem into two stages: designing or learning visual features, and prediction. More recent deep lipreading approaches are end-to-end trainable (Wand et al., 2016; Chung & Zisserman, 2016a). However, exi...
LipNet: End-to-End Sentence-level Lipreading
2,016
http://arxiv.org/pdf/1611.01599v2
Title LipNet EndtoEnd Sentencelevel Lipreading Summary Lipreading task decoding text movement speaker mouth Traditional approach separated problem two stage designing learning visual feature prediction recent deep lipreading approach endtoend trainable Wand et al 2016 Chung Zisserman 2016a However existing work model t...
[0.0153832221403718, 0.01479147095233202, 0.02952849678695202, 0.05985181778669357, 0.023813242092728615, 0.0009587219101376832, 0.040750958025455475, -0.009336501359939575, -0.037979017943143845, 0.012710895389318466, -0.043451547622680664, 0.002306808717548847, 0.041850510984659195, 0.03284887224435806, 0.03728131949...
1,925
1,925
['Abhinav Thanda', 'Shankar M Venkatesan']
1611.02879v1
In this work, we propose a training algorithm for an audio-visual automatic speech recognition (AV-ASR) system using deep recurrent neural network (RNN).First, we train a deep RNN acoustic model with a Connectionist Temporal Classification (CTC) objective function. The frame labels obtained from the acoustic model are ...
Audio Visual Speech Recognition using Deep Recurrent Neural Networks
2,016
http://arxiv.org/pdf/1611.02879v1
Title Audio Visual Speech Recognition using Deep Recurrent Neural Networks Summary work propose training algorithm audiovisual automatic speech recognition AVASR system using deep recurrent neural network RNNFirst train deep RNN acoustic model Connectionist Temporal Classification CTC objective function frame label obt...
[-0.013347885571420193, -0.004991111345589161, 0.026523463428020477, 0.05694941058754921, 0.029855405911803246, -0.01577630080282688, 0.031258367002010345, -0.02945016697049141, -0.042567070573568344, 0.005845132749527693, -0.08016761392354965, -0.014463221654295921, 0.07093142718076706, 0.04534507915377617, 0.01084045...
1,926
1,926
['Jie Mei', 'Aminul Islam', 'Yajing Wu', "Abidalrahman Moh'd", 'Evangelos E. Milios']
1611.06950v1
The accuracy of Optical Character Recognition (OCR) is crucial to the success of subsequent applications used in text analyzing pipeline. Recent models of OCR post-processing significantly improve the quality of OCR-generated text, but are still prone to suggest correction candidates from limited observations while ins...
Statistical Learning for OCR Text Correction
2,016
http://arxiv.org/pdf/1611.06950v1
Title Statistical Learning OCR Text Correction Summary accuracy Optical Character Recognition OCR crucial success subsequent application used text analyzing pipeline Recent model OCR postprocessing significantly improve quality OCRgenerated text still prone suggest correction candidate limited observation insufficientl...
[0.012530646286904812, 0.016086973249912262, 0.03476725146174431, 0.058697376400232315, -0.05679081752896309, -0.018245326355099678, 0.022005200386047363, 0.06314581632614136, -0.009433713741600513, -0.04421316459774971, 0.029805177822709084, 0.05064011365175247, 0.09690473973751068, 0.02296479046344757, -0.05355256050...
1,927
1,927
['Zhe Gan', 'Chuang Gan', 'Xiaodong He', 'Yunchen Pu', 'Kenneth Tran', 'Jianfeng Gao', 'Lawrence Carin', 'Li Deng']
1611.08002v2
A Semantic Compositional Network (SCN) is developed for image captioning, in which semantic concepts (i.e., tags) are detected from the image, and the probability of each tag is used to compose the parameters in a long short-term memory (LSTM) network. The SCN extends each weight matrix of the LSTM to an ensemble of ta...
Semantic Compositional Networks for Visual Captioning
2,016
http://arxiv.org/pdf/1611.08002v2
Title Semantic Compositional Networks Visual Captioning Summary Semantic Compositional Network SCN developed image captioning semantic concept ie tag detected image probability tag used compose parameter long shortterm memory LSTM network SCN extends weight matrix LSTM ensemble tagdependent weight matrix degree member ...
[0.03757045790553093, 0.007851365022361279, 0.010910444892942905, 0.039530184119939804, -0.010543935000896454, 0.024578021839261055, 0.01089874655008316, -0.003615268040448427, -0.02385123446583748, -0.05445566400885582, 0.032427068799734116, 0.012191839516162872, 0.01608513481914997, 0.06343036890029907, -0.0097362790...
1,928
1,928
['Junhua Mao', 'Jiajing Xu', 'Yushi Jing', 'Alan Yuille']
1611.08321v1
In this paper, we focus on training and evaluating effective word embeddings with both text and visual information. More specifically, we introduce a large-scale dataset with 300 million sentences describing over 40 million images crawled and downloaded from publicly available Pins (i.e. an image with sentence descript...
Training and Evaluating Multimodal Word Embeddings with Large-scale Web Annotated Images
2,016
http://arxiv.org/pdf/1611.08321v1
Title Training Evaluating Multimodal Word Embeddings Largescale Web Annotated Images Summary paper focus training evaluating effective word embeddings text visual information specifically introduce largescale dataset 300 million sentence describing 40 million image crawled downloaded publicly available Pins ie image se...
[0.06229444220662117, 0.04569285362958908, 0.010145047679543495, 0.08391416817903519, -0.009813059121370316, -0.006584987975656986, 0.006694173440337181, -0.0004635954392142594, -0.022023813799023628, -0.05768460035324097, -0.026711495593190193, 0.014367236755788326, -0.009645265527069569, 0.02174447849392891, 0.027005...
1,929
1,929
['Satish Palaniappan', 'Ronojoy Adhikari']
1702.00523v1
Standardized corpora of undeciphered scripts, a necessary starting point for computational epigraphy, requires laborious human effort for their preparation from raw archaeological records. Automating this process through machine learning algorithms can be of significant aid to epigraphical research. Here, we take the f...
Deep Learning the Indus Script
2,017
http://arxiv.org/pdf/1702.00523v1
Title Deep Learning Indus Script Summary Standardized corpus undeciphered script necessary starting point computational epigraphy requires laborious human effort preparation raw archaeological record Automating process machine learning algorithm significant aid epigraphical research take first step direction present de...
[0.009258639067411423, 0.038935285061597824, -0.03189752623438835, 0.055631041526794434, -0.032869454473257065, 0.025025783106684685, 0.02924126759171486, 0.06124240905046463, 0.032286226749420166, -0.007984030991792679, 0.06261150538921356, 0.014245666563510895, 0.05323003605008125, 0.04172979295253754, 0.020523518323...
1,930
1,930
['Yao-Hung Hubert Tsai', 'Liang-Kang Huang', 'Ruslan Salakhutdinov']
1703.05908v2
Many of the existing methods for learning joint embedding of images and text use only supervised information from paired images and its textual attributes. Taking advantage of the recent success of unsupervised learning in deep neural networks, we propose an end-to-end learning framework that is able to extract more ro...
Learning Robust Visual-Semantic Embeddings
2,017
http://arxiv.org/pdf/1703.05908v2
Title Learning Robust VisualSemantic Embeddings Summary Many existing method learning joint embedding image text use supervised information paired image textual attribute Taking advantage recent success unsupervised learning deep neural network propose endtoend learning framework able extract robust multimodal represen...
[-0.014256333000957966, 0.029021963477134705, -0.012860733084380627, 0.04951856657862663, -0.00460374029353261, 0.009860038757324219, 0.004748955834656954, 0.0072862827219069, -0.005552264861762524, -0.0640937015414238, -0.0021317435894161463, 0.03527496010065079, 0.011859242804348469, 0.034903738647699356, 0.039595022...
1,931
1,931
['Justin Johnson', 'Bharath Hariharan', 'Laurens van der Maaten', 'Judy Hoffman', 'Li Fei-Fei', 'C. Lawrence Zitnick', 'Ross Girshick']
1705.03633v1
Existing methods for visual reasoning attempt to directly map inputs to outputs using black-box architectures without explicitly modeling the underlying reasoning processes. As a result, these black-box models often learn to exploit biases in the data rather than learning to perform visual reasoning. Inspired by module...
Inferring and Executing Programs for Visual Reasoning
2,017
http://arxiv.org/pdf/1705.03633v1
Title Inferring Executing Programs Visual Reasoning Summary Existing method visual reasoning attempt directly map input output using blackbox architecture without explicitly modeling underlying reasoning process result blackbox model often learn exploit bias data rather learning perform visual reasoning Inspired module...
[-0.005308021791279316, 0.0016802240861579776, -0.042408134788274765, 0.029550572857260704, -0.02508360520005226, 0.0011384106473997235, -0.013676638714969158, 0.018334591761231422, -0.0004429493856150657, -0.028254078701138496, 0.03561275079846382, 0.09080764651298523, 0.010369697585701942, 0.07877660542726517, 0.0440...
1,932
1,932
['Karol Kurach', 'Sylvain Gelly', 'Michal Jastrzebski', 'Philip Haeusser', 'Olivier Teytaud', 'Damien Vincent', 'Olivier Bousquet']
1705.08386v2
Generic text embeddings are successfully used in a variety of tasks. However, they are often learnt by capturing the co-occurrence structure from pure text corpora, resulting in limitations of their ability to generalize. In this paper, we explore models that incorporate visual information into the text representation....
Better Text Understanding Through Image-To-Text Transfer
2,017
http://arxiv.org/pdf/1705.08386v2
Title Better Text Understanding ImageToText Transfer Summary Generic text embeddings successfully used variety task However often learnt capturing cooccurrence structure pure text corpus resulting limitation ability generalize paper explore model incorporate visual information text representation Based comprehensive ab...
[0.011716960929334164, 0.018181290477514267, -0.004495323169976473, 0.0658583790063858, -0.040386997163295746, 0.015920210629701614, 0.004607372917234898, 0.05799100175499916, 0.0020683908369392157, -0.05438396707177162, -0.023224495351314545, -0.016682453453540802, 0.022892441600561142, 0.07711713761091232, 0.01502747...
1,933
1,933
['Serhii Havrylov', 'Ivan Titov']
1705.11192v2
Learning to communicate through interaction, rather than relying on explicit supervision, is often considered a prerequisite for developing a general AI. We study a setting where two agents engage in playing a referential game and, from scratch, develop a communication protocol necessary to succeed in this game. Unlike...
Emergence of Language with Multi-agent Games: Learning to Communicate with Sequences of Symbols
2,017
http://arxiv.org/pdf/1705.11192v2
Title Emergence Language Multiagent Games Learning Communicate Sequences Symbols Summary Learning communicate interaction rather relying explicit supervision often considered prerequisite developing general AI study setting two agent engage playing referential game scratch develop communication protocol necessary succe...
[0.05379199609160423, 0.017209434881806374, -0.034755539149045944, 0.010731332935392857, -0.06160767003893852, -0.00632530776783824, -0.0028862354811280966, 0.01890427991747856, -0.013685098849236965, -0.016145281493663788, 0.024173203855752945, -0.003146185539662838, 0.04240715131163597, 0.07258296757936478, 0.0240452...
1,934
1,934
['Harm de Vries', 'Florian Strub', 'Jérémie Mary', 'Hugo Larochelle', 'Olivier Pietquin', 'Aaron Courville']
1707.00683v3
It is commonly assumed that language refers to high-level visual concepts while leaving low-level visual processing unaffected. This view dominates the current literature in computational models for language-vision tasks, where visual and linguistic input are mostly processed independently before being fused into a sin...
Modulating early visual processing by language
2,017
http://arxiv.org/pdf/1707.00683v3
Title Modulating early visual processing language Summary commonly assumed language refers highlevel visual concept leaving lowlevel visual processing unaffected view dominates current literature computational model languagevision task visual linguistic input mostly processed independently fused single representation p...
[0.02513597160577774, 0.031430646777153015, -0.028609156608581543, 0.02403467521071434, 0.01029114332050085, -0.003399238456040621, -0.00980179850012064, 0.016650982201099396, -0.03707151114940643, -0.04293356463313103, -0.008771526627242565, 0.0042634685523808, 0.0008119286503642797, 0.04725205898284912, 0.03695069998...
1,935
1,935
['Fartash Faghri', 'David J. Fleet', 'Jamie Ryan Kiros', 'Sanja Fidler']
1707.05612v2
We present a new technique for learning visual-semantic embeddings for cross-modal retrieval. Inspired by the use of hard negatives in structured prediction, and ranking loss functions used in retrieval, we introduce a simple change to common loss functions used to learn multi-modal embeddings. That, combined with fine...
VSE++: Improving Visual-Semantic Embeddings with Hard Negatives
2,017
http://arxiv.org/pdf/1707.05612v2
Title VSE Improving VisualSemantic Embeddings Hard Negatives Summary present new technique learning visualsemantic embeddings crossmodal retrieval Inspired use hard negative structured prediction ranking loss function used retrieval introduce simple change common loss function used learn multimodal embeddings combined ...
[0.0036112479865550995, 0.020161984488368034, -0.006348464637994766, 0.044735971838235855, 0.01292404718697071, 0.003998636733740568, -0.02723362296819687, 0.0354233980178833, 0.010819043032824993, -0.04138565808534622, -0.037703365087509155, 0.005576586816459894, -0.026687005534768105, 0.05209432542324066, 0.032528061...
1,936
1,936
['Chuang Gan', 'Yandong Li', 'Haoxiang Li', 'Chen Sun', 'Boqing Gong']
1708.04686v1
Rich and dense human labeled datasets are among the main enabling factors for the recent advance on vision-language understanding. Many seemingly distant annotations (e.g., semantic segmentation and visual question answering (VQA)) are inherently connected in that they reveal different levels and perspectives of human ...
VQS: Linking Segmentations to Questions and Answers for Supervised Attention in VQA and Question-Focused Semantic Segmentation
2,017
http://arxiv.org/pdf/1708.04686v1
Title VQS Linking Segmentations Questions Answers Supervised Attention VQA QuestionFocused Semantic Segmentation Summary Rich dense human labeled datasets among main enabling factor recent advance visionlanguage understanding Many seemingly distant annotation eg semantic segmentation visual question answering VQA inher...
[0.046043314039707184, 0.009113709442317486, -0.013996644876897335, 0.07710210978984833, 0.004497261252254248, 0.019251450896263123, 0.032988935708999634, -0.013253811746835709, 0.0027865376323461533, -0.0444481261074543, -0.009124350734055042, 0.016219450160861015, -0.00499297771602869, 0.018901048228144646, 0.0361367...
1,937
1,937
['Yang Xian', 'Yingli Tian']
1709.05038v1
In this paper, a self-guiding multimodal LSTM (sg-LSTM) image captioning model is proposed to handle uncontrolled imbalanced real-world image-sentence dataset. We collect FlickrNYC dataset from Flickr as our testbed with 306,165 images and the original text descriptions uploaded by the users are utilized as the ground ...
Self-Guiding Multimodal LSTM - when we do not have a perfect training dataset for image captioning
2,017
http://arxiv.org/pdf/1709.05038v1
Title SelfGuiding Multimodal LSTM perfect training dataset image captioning Summary paper selfguiding multimodal LSTM sgLSTM image captioning model proposed handle uncontrolled imbalanced realworld imagesentence dataset collect FlickrNYC dataset Flickr testbed 306165 image original text description uploaded user utiliz...
[0.0609104186296463, 0.017605571076273918, 0.020481960847973824, 0.05718410015106201, -0.022822491824626923, 0.01052162516862154, 0.027681224048137665, -0.011246905662119389, -0.018151406198740005, -0.08095657080411911, -0.043236251920461655, -0.014410714618861675, -0.007032691966742277, 0.04246456176042557, 0.00617270...
1,938
1,938
['Xu Sun', 'Bingzhen Wei', 'Xuancheng Ren', 'Shuming Ma']
1710.10393v1
We propose a method, called Label Embedding Network, which can learn label representation (label embedding) during the training process of deep networks. With the proposed method, the label embedding is adaptively and automatically learned through back propagation. The original one-hot represented loss function is conv...
Label Embedding Network: Learning Label Representation for Soft Training of Deep Networks
2,017
http://arxiv.org/pdf/1710.10393v1
Title Label Embedding Network Learning Label Representation Soft Training Deep Networks Summary propose method called Label Embedding Network learn label representation label embedding training process deep network proposed method label embedding adaptively automatically learned back propagation original onehot represe...
[-0.0041319807060062885, 0.013407545164227486, -0.005647705867886543, 0.05988280847668648, 0.018408987671136856, -0.006726698484271765, 0.03320416063070297, -0.015212559141218662, 0.007989793084561825, -0.012567042373120785, -0.010755392722785473, 0.04677826166152954, -0.032197754830121994, 0.05329598858952522, 0.02842...
1,939
1,939
['Jianbo Chen', 'Yelong Shen', 'Jianfeng Gao', 'Jingjing Liu', 'Xiaodong Liu']
1711.06288v1
We investigate the problem of Language-Based Image Editing (LBIE) in this work. Given a source image and a natural language description, we want to generate a target image by editing the source im- age based on the description. We propose a generic modeling framework for two sub-tasks of LBIE: language-based image segm...
Language-Based Image Editing with Recurrent Attentive Models
2,017
http://arxiv.org/pdf/1711.06288v1
Title LanguageBased Image Editing Recurrent Attentive Models Summary investigate problem LanguageBased Image Editing LBIE work Given source image natural language description want generate target image editing source im age based description propose generic modeling framework two subtasks LBIE languagebased image segme...
[0.054590508341789246, 0.015464499592781067, 0.018978577107191086, 0.07335062325000763, -0.009570534341037273, -0.0005884341662749648, 0.037497133016586304, 0.007010838016867638, -0.06078018248081207, -0.05507772043347359, -0.02914297766983509, -0.05051392689347267, 0.019017856568098068, 0.05596606805920601, 0.01104517...
1,940
1,940
['Ishan Misra', 'Ross Girshick', 'Rob Fergus', 'Martial Hebert', 'Abhinav Gupta', 'Laurens van der Maaten']
1712.01238v1
We introduce an interactive learning framework for the development and testing of intelligent visual systems, called learning-by-asking (LBA). We explore LBA in context of the Visual Question Answering (VQA) task. LBA differs from standard VQA training in that most questions are not observed during training time, and t...
Learning by Asking Questions
2,017
http://arxiv.org/pdf/1712.01238v1
Title Learning Asking Questions Summary introduce interactive learning framework development testing intelligent visual system called learningbyasking LBA explore LBA context Visual Question Answering VQA task LBA differs standard VQA training question observed training time learner must ask question want answer Thus L...
[0.04647989198565483, -0.03832748532295227, -0.0320994071662426, 0.0533619225025177, -0.022607797756791115, 0.042575374245643616, -0.013398385606706142, 0.027958247810602188, 0.03898663446307182, -0.021292632445693016, 0.004242136143147945, 0.03137976676225662, -0.043840523809194565, 0.04898945614695549, 0.040218207985...
1,941
1,941
['Kenneth Leidal', 'David Harwath', 'James Glass']
1712.03897v1
In this paper, we explore the unsupervised learning of a semantic embedding space for co-occurring sensory inputs. Specifically, we focus on the task of learning a semantic vector space for both spoken and handwritten digits using the TIDIGITs and MNIST datasets. Current techniques encode image and audio/textual inputs...
Learning Modality-Invariant Representations for Speech and Images
2,017
http://arxiv.org/pdf/1712.03897v1
Title Learning ModalityInvariant Representations Speech Images Summary paper explore unsupervised learning semantic embedding space cooccurring sensory input Specifically focus task learning semantic vector space spoken handwritten digit using TIDIGITs MNIST datasets Current technique encode image audiotextual input di...
[-0.03603636100888252, 0.047251325100660324, -0.011598740704357624, 0.05210009217262268, -0.013642596080899239, 0.008318580687046051, 0.03387545049190521, -0.002447296865284443, -0.06035871058702469, -0.04861799627542496, -0.07990363240242004, 0.02827667072415352, 0.004196727182716131, 0.044094718992710114, 0.032089594...
1,942
1,942
['Jason Xie', 'Tingwen Bao']
1712.06682v1
Generating novel pairs of image and text is a problem that combines computer vision and natural language processing. In this paper, we present strategies for generating novel image and caption pairs based on existing captioning datasets. The model takes advantage of recent advances in generative adversarial networks an...
Synthesizing Novel Pairs of Image and Text
2,017
http://arxiv.org/pdf/1712.06682v1
Title Synthesizing Novel Pairs Image Text Summary Generating novel pair image text problem combine computer vision natural language processing paper present strategy generating novel image caption pair based existing captioning datasets model take advantage recent advance generative adversarial network sequencetosequen...
[0.049424827098846436, 0.059592943638563156, -0.008910011500120163, 0.05295021831989288, -0.02145352028310299, -0.007907837629318237, 0.025088030844926834, -0.00743473507463932, 0.029772890731692314, -0.02822738327085972, -0.010431253351271152, -0.039709482342004776, 0.006968350615352392, 0.09035183489322662, 0.0197790...
1,943
1,943
['Pelin Dogan', 'Boyang Li', 'Leonid Sigal', 'Markus Gross']
1803.00057v1
The alignment of heterogeneous sequential data (video to text) is an important and challenging problem. Standard techniques for such alignment, including Dynamic Time Warping (DTW) and Conditional Random Fields (CRFs), suffer from inherent drawbacks. Mainly, the Markov assumption implies that, given the immediate past,...
LSTM stack-based Neural Multi-sequence Alignment TeCHnique (NeuMATCH)
2,018
http://arxiv.org/pdf/1803.00057v1
Title LSTM stackbased Neural Multisequence Alignment TeCHnique NeuMATCH Summary alignment heterogeneous sequential data video text important challenging problem Standard technique alignment including Dynamic Time Warping DTW Conditional Random Fields CRFs suffer inherent drawback Mainly Markov assumption implies given ...
[-0.003863204037770629, 0.06979136914014816, 0.019424989819526672, 0.017292406409978867, -0.026835227385163307, -0.0004638778336811811, 0.0285099558532238, 0.01350557804107666, -0.02321733348071575, 0.015686403959989548, -0.004273180849850178, -0.05907075107097626, 0.05682188272476196, -0.0010819692397490144, 0.0119762...
1,944
1,944
['Amos Storkey']
1106.4509v1
Prediction markets show considerable promise for developing flexible mechanisms for machine learning. Here, machine learning markets for multivariate systems are defined, and a utility-based framework is established for their analysis. This differs from the usual approach of defining static betting functions. It is sho...
Machine Learning Markets
2,011
http://arxiv.org/pdf/1106.4509v1
Title Machine Learning Markets Summary Prediction market show considerable promise developing flexible mechanism machine learning machine learning market multivariate system defined utilitybased framework established analysis differs usual approach defining static betting function shown market implement model combinati...
[0.016158202663064003, 0.021132757887244225, -0.05616506189107895, -0.07620953768491745, -0.0015008344780653715, -0.016355184838175774, 0.08429225534200668, -0.005561568774282932, 0.0011157892877236009, -0.0017223723698407412, 0.04048127681016922, 0.04981696978211403, -0.013685200363397598, 0.13280357420444489, 0.04888...
1,945
1,945
['M. Sanz', 'L. Lamata', 'E. Solano']
1709.07808v1
We propose a method to build quantum memristors in quantum photonic platforms. We firstly design an effective beam splitter, which is tunable in real-time, by means of a Mach-Zehnder-type array with two equal 50:50 beam splitters and a tunable retarder, which allows us to control its reflectivity. Then, we show that th...
Quantum Memristors in Quantum Photonics
2,017
http://arxiv.org/pdf/1709.07808v1
Title Quantum Memristors Quantum Photonics Summary propose method build quantum memristors quantum photonic platform firstly design effective beam splitter tunable realtime mean MachZehndertype array two equal 5050 beam splitter tunable retarder allows u control reflectivity show tunable beam splitter equipped weak mea...
[-0.04002555459737778, 0.0012817558599635959, -0.05555417016148567, 0.055808279663324356, -0.030166607350111008, -0.004903675988316536, -0.02860843949019909, -0.016976123675704002, 0.03496330976486206, -0.01829426921904087, -0.04458516836166382, -0.015128526836633682, -0.061451323330402374, 0.08596628904342651, 0.02585...
1,946
1,946
['Lukas Cavigelli', 'Dominic Bernath', 'Michele Magno', 'Luca Benini']
1611.03130v1
Detecting and classifying targets in video streams from surveillance cameras is a cumbersome, error-prone and expensive task. Often, the incurred costs are prohibitive for real-time monitoring. This leads to data being stored locally or transmitted to a central storage site for post-incident examination. The required c...
Computationally Efficient Target Classification in Multispectral Image Data with Deep Neural Networks
2,016
http://arxiv.org/pdf/1611.03130v1
Title Computationally Efficient Target Classification Multispectral Image Data Deep Neural Networks Summary Detecting classifying target video stream surveillance camera cumbersome errorprone expensive task Often incurred cost prohibitive realtime monitoring lead data stored locally transmitted central storage site pos...
[0.006198809016495943, 0.03297511488199234, 0.012050468474626541, 0.03435709327459335, -0.017081232741475105, -0.03242902457714081, 0.07603206485509872, -0.04760618880391121, -0.05787106230854988, -0.0533931627869606, 0.02650459297001362, 0.030627934262156487, 0.026134224608540535, 0.06805604696273804, 0.01051885914057...
1,947
1,947
['Arunkumar Byravan', 'Felix Leeb', 'Franziska Meier', 'Dieter Fox']
1710.00489v1
In this work, we present an approach to deep visuomotor control using structured deep dynamics models. Our deep dynamics model, a variant of SE3-Nets, learns a low-dimensional pose embedding for visuomotor control via an encoder-decoder structure. Unlike prior work, our dynamics model is structured: given an input scen...
SE3-Pose-Nets: Structured Deep Dynamics Models for Visuomotor Planning and Control
2,017
http://arxiv.org/pdf/1710.00489v1
Title SE3PoseNets Structured Deep Dynamics Models Visuomotor Planning Control Summary work present approach deep visuomotor control using structured deep dynamic model deep dynamic model variant SE3Nets learns lowdimensional pose embedding visuomotor control via encoderdecoder structure Unlike prior work dynamic model ...
[-0.014479981735348701, -0.029330018907785416, 0.009508280083537102, 0.013032814487814903, 0.014198262244462967, 0.007458385545760393, -0.007270668633282185, -0.026687394827604294, 0.00043656857451424, -0.024139072746038437, -0.035000935196876526, 0.01515317615121603, -0.01362442597746849, 0.07919039577245712, 0.024073...
1,948
1,948
['Wellington Pinheiro dos Santos', 'Francisco Marcos de Assis', 'Ricardo Emmanuel de Souza', 'Priscilla B. Mendes', 'Henrique S. S. Monteiro', 'Havana Diogo Alves']
1712.01694v1
The materialist dialectical method is a philosophical investigative method to analyze aspects of reality. These aspects are viewed as complex processes composed by basic units named poles, which interact with each other. Dialectics has experienced considerable progress in the 19th century, with Hegel's dialectics and, ...
Fuzzy-Based Dialectical Non-Supervised Image Classification and Clustering
2,017
http://arxiv.org/pdf/1712.01694v1
Title FuzzyBased Dialectical NonSupervised Image Classification Clustering Summary materialist dialectical method philosophical investigative method analyze aspect reality aspect viewed complex process composed basic unit named pole interact Dialectics experienced considerable progress 19th century Hegels dialectic 20t...
[-0.009561926126480103, 0.06413190066814423, -0.03887389227747917, 0.023286066949367523, 0.024578867480158806, 0.028884541243314743, 0.02170419879257679, 0.07942850142717361, 0.008286935277283192, -0.025443218648433685, 0.03229217603802681, 0.012474185787141323, 0.008599602617323399, -0.03127027675509453, -0.0143457371...
1,949
1,949
['Mohammadreza Zolfaghari', 'Gabriel L. Oliveira', 'Nima Sedaghat', 'Thomas Brox']
1704.00616v2
General human action recognition requires understanding of various visual cues. In this paper, we propose a network architecture that computes and integrates the most important visual cues for action recognition: pose, motion, and the raw images. For the integration, we introduce a Markov chain model which adds cues su...
Chained Multi-stream Networks Exploiting Pose, Motion, and Appearance for Action Classification and Detection
2,017
http://arxiv.org/pdf/1704.00616v2
Title Chained Multistream Networks Exploiting Pose Motion Appearance Action Classification Detection Summary General human action recognition requires understanding various visual cue paper propose network architecture computes integrates important visual cue action recognition pose motion raw image integration introdu...
[-0.028757261112332344, 0.012512513436377048, -0.01760033331811428, 0.03246408700942993, 0.002898276085034013, 0.020851563662290573, 0.04290839284658432, -0.006174824200570583, 0.013563087210059166, -0.0508565753698349, -0.01604834385216236, -0.04678159952163696, 0.0011554558295756578, 0.05488160252571106, 0.0293436199...
1,950
1,950
['Filipe Rolim Cordeiro', 'Wellington Pinheiro dos Santos', 'Abel Guilhermino da Silva Filho']
1801.01443v1
According to the World Health Organization, breast cancer is the most common form of cancer in women. It is the second leading cause of death among women round the world, becoming the most fatal form of cancer. Mammographic image segmentation is a fundamental task to support image analysis and diagnosis, taking into ac...
A semi-supervised fuzzy GrowCut algorithm to segment and classify regions of interest of mammographic images
2,017
http://arxiv.org/pdf/1801.01443v1
Title semisupervised fuzzy GrowCut algorithm segment classify region interest mammographic image Summary According World Health Organization breast cancer common form cancer woman second leading cause death among woman round world becoming fatal form cancer Mammographic image segmentation fundamental task support image...
[0.02392052672803402, 0.01936122216284275, -0.030704475939273834, -0.0041809831745922565, -0.013607854954898357, 0.031180711463093758, 0.01179508212953806, 0.014250610023736954, 0.0024159757886081934, 0.018662793561816216, 0.05406421795487404, -0.0059408457018435, 0.019317612051963806, 0.05267723277211189, 0.0009894749...
1,951
1,951
['Peng Sun', 'Mark D. Reid', 'Jie Zhou']
1110.3907v3
This paper presents an improvement to model learning when using multi-class LogitBoost for classification. Motivated by the statistical view, LogitBoost can be seen as additive tree regression. Two important factors in this setting are: 1) coupled classifier output due to a sum-to-zero constraint, and 2) the dense Hess...
AOSO-LogitBoost: Adaptive One-Vs-One LogitBoost for Multi-Class Problem
2,011
http://arxiv.org/pdf/1110.3907v3
Title AOSOLogitBoost Adaptive OneVsOne LogitBoost MultiClass Problem Summary paper present improvement model learning using multiclass LogitBoost classification Motivated statistical view LogitBoost seen additive tree regression Two important factor setting 1 coupled classifier output due sumtozero constraint 2 dense H...
[-0.03296606242656708, 0.05653146654367447, -0.006270922254770994, -0.019986161962151527, 0.03531356528401375, 0.009633042849600315, 0.007502211257815361, 0.024965625256299973, -0.013554667122662067, -0.028252027928829193, 0.011459332890808582, -0.01497524231672287, 0.020197397097945213, 0.014676141552627087, -0.036190...
1,952
1,952
['Andreas C. Damianou', 'Michalis K. Titsias', 'Neil D. Lawrence']
1107.4985v1
High dimensional time series are endemic in applications of machine learning such as robotics (sensor data), computational biology (gene expression data), vision (video sequences) and graphics (motion capture data). Practical nonlinear probabilistic approaches to this data are required. In this paper we introduce the v...
Variational Gaussian Process Dynamical Systems
2,011
http://arxiv.org/pdf/1107.4985v1
Title Variational Gaussian Process Dynamical Systems Summary High dimensional time series endemic application machine learning robotics sensor data computational biology gene expression data vision video sequence graphic motion capture data Practical nonlinear probabilistic approach data required paper introduce variat...
[-0.03083920292556286, 0.06631612777709961, -0.031006187200546265, -0.0202065147459507, 0.0013108662096783519, 0.01850186660885811, 0.005673545878380537, -0.003497848054394126, -0.06549949198961258, 0.019826626405119896, 0.06181999295949936, -0.02682526223361492, 0.02342562936246395, 0.10817407071590424, 0.010855323635...
1,953
1,953
['Khoat Than', 'Tu Bao Ho']
1210.7053v2
Inference is an integral part of probabilistic topic models, but is often non-trivial to derive an efficient algorithm for a specific model. It is even much more challenging when we want to find a fast inference algorithm which always yields sparse latent representations of documents. In this article, we introduce a si...
Managing sparsity, time, and quality of inference in topic models
2,012
http://arxiv.org/pdf/1210.7053v2
Title Managing sparsity time quality inference topic model Summary Inference integral part probabilistic topic model often nontrivial derive efficient algorithm specific model even much challenging want find fast inference algorithm always yield sparse latent representation document article introduce simple framework i...
[0.026217272505164146, 0.04010126367211342, -0.0006178987096063793, 0.023340871557593346, -0.03273701295256615, -0.02975769340991974, 0.011913889087736607, 0.0020416956394910812, -0.07666625082492828, -0.045265667140483856, 0.03134548291563988, 0.018887151032686234, 0.016947319731116295, 0.03883935511112213, 0.00902461...
1,954
1,954
['Dinesh Jayaraman', 'Kristen Grauman']
1505.02206v2
Understanding how images of objects and scenes behave in response to specific ego-motions is a crucial aspect of proper visual development, yet existing visual learning methods are conspicuously disconnected from the physical source of their images. We propose to exploit proprioceptive motor signals to provide unsuperv...
Learning image representations tied to ego-motion
2,015
http://arxiv.org/pdf/1505.02206v2
Title Learning image representation tied egomotion Summary Understanding image object scene behave response specific egomotions crucial aspect proper visual development yet existing visual learning method conspicuously disconnected physical source image propose exploit proprioceptive motor signal provide unsupervised r...
[0.004861196503043175, 0.05312051624059677, -0.013334537856280804, 0.060094740241765976, -0.014360965229570866, -0.00657928129658103, 0.031503353267908096, -0.013995381072163582, -0.013675793074071407, -0.05009421706199646, -0.034869443625211716, 0.04285014793276787, 0.00045034135109744966, 0.018662504851818085, 0.0559...
1,955
1,955
['Han Zhang', 'Tao Xu', 'Hongsheng Li', 'Shaoting Zhang', 'Xiaogang Wang', 'Xiaolei Huang', 'Dimitris Metaxas']
1612.03242v2
Synthesizing high-quality images from text descriptions is a challenging problem in computer vision and has many practical applications. Samples generated by existing text-to-image approaches can roughly reflect the meaning of the given descriptions, but they fail to contain necessary details and vivid object parts. In...
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks
2,016
http://arxiv.org/pdf/1612.03242v2
Title StackGAN Text Photorealistic Image Synthesis Stacked Generative Adversarial Networks Summary Synthesizing highquality image text description challenging problem computer vision many practical application Samples generated existing texttoimage approach roughly reflect meaning given description fail contain necessa...
[0.004407680127769709, 0.0789143368601799, 0.018100561574101448, 0.03592776879668236, -0.0011214129626750946, -0.02349100261926651, 0.004202753771096468, 0.008034833706915379, -0.050943028181791306, 0.006091931369155645, 0.007521857973188162, -0.035865988582372665, -0.0016414282144978642, 0.04541533440351486, 0.0373324...
1,956
1,956
['Mehdi S. M. Sajjadi', 'Bernhard Schölkopf', 'Michael Hirsch']
1612.07919v2
Single image super-resolution is the task of inferring a high-resolution image from a single low-resolution input. Traditionally, the performance of algorithms for this task is measured using pixel-wise reconstruction measures such as peak signal-to-noise ratio (PSNR) which have been shown to correlate poorly with the ...
EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis
2,016
http://arxiv.org/pdf/1612.07919v2
Title EnhanceNet Single Image SuperResolution Automated Texture Synthesis Summary Single image superresolution task inferring highresolution image single lowresolution input Traditionally performance algorithm task measured using pixelwise reconstruction measure peak signaltonoise ratio PSNR shown correlate poorly huma...
[0.017575522884726524, 0.07690668851137161, 0.0029536853544414043, 0.06364759802818298, 0.008040985092520714, -0.0515047088265419, -0.010118509642779827, 0.002637502271682024, -0.07499141991138458, 0.05689803510904312, 0.025371678173542023, 0.029098451137542725, -0.0034516314044594765, 0.013382057659327984, 0.049479138...
1,957
1,957
['Vikash K. Mansinghka', 'Tejas D. Kulkarni', 'Yura N. Perov', 'Joshua B. Tenenbaum']
1307.0060v1
The idea of computer vision as the Bayesian inverse problem to computer graphics has a long history and an appealing elegance, but it has proved difficult to directly implement. Instead, most vision tasks are approached via complex bottom-up processing pipelines. Here we show that it is possible to write short, simple ...
Approximate Bayesian Image Interpretation using Generative Probabilistic Graphics Programs
2,013
http://arxiv.org/pdf/1307.0060v1
Title Approximate Bayesian Image Interpretation using Generative Probabilistic Graphics Programs Summary idea computer vision Bayesian inverse problem computer graphic long history appealing elegance proved difficult directly implement Instead vision task approached via complex bottomup processing pipeline show possibl...
[-0.011626800522208214, 0.05189124122262001, 0.007916132919490337, 0.04802766069769859, -0.04256373271346092, 0.015114221721887589, -0.0033600821625441313, 0.039497386664152145, -0.054562292993068695, 0.002229855628684163, 0.04348060488700867, 0.012416288256645203, 0.05728089064359665, 0.071538545191288, 0.005507067311...
1,958
1,958
['Tejas D. Kulkarni', 'Vikash K. Mansinghka', 'Pushmeet Kohli', 'Joshua B. Tenenbaum']
1407.1339v1
Recently, multiple formulations of vision problems as probabilistic inversions of generative models based on computer graphics have been proposed. However, applications to 3D perception from natural images have focused on low-dimensional latent scenes, due to challenges in both modeling and inference. Accounting for th...
Inverse Graphics with Probabilistic CAD Models
2,014
http://arxiv.org/pdf/1407.1339v1
Title Inverse Graphics Probabilistic CAD Models Summary Recently multiple formulation vision problem probabilistic inversion generative model based computer graphic proposed However application 3D perception natural image focused lowdimensional latent scene due challenge modeling inference Accounting enormous variabili...
[0.012258650735020638, 0.01590992696583271, 0.0070972018875181675, 0.03177201375365257, -0.021831922233104706, -0.008755332790315151, -0.022285103797912598, -0.01631012000143528, -0.0594063438475132, 0.005801831372082233, 0.02545652538537979, 0.004951040260493755, 0.0337098129093647, 0.03812510892748833, 0.049214668571...
1,959
1,959
['Harris V. Georgiou']
1410.7100v1
Functional Magnetic Resonance Imaging (fMRI) is a powerful non-invasive tool for localizing and analyzing brain activity. This study focuses on one very important aspect of the functional properties of human brain, specifically the estimation of the level of parallelism when performing complex cognitive tasks. Using fM...
Estimating the intrinsic dimension in fMRI space via dataset fractal analysis - Counting the `cpu cores' of the human brain
2,014
http://arxiv.org/pdf/1410.7100v1
Title Estimating intrinsic dimension fMRI space via dataset fractal analysis Counting cpu core human brain Summary Functional Magnetic Resonance Imaging fMRI powerful noninvasive tool localizing analyzing brain activity study focus one important aspect functional property human brain specifically estimation level paral...
[-0.04860258474946022, -0.005726220551878214, -0.052349288016557693, -0.005320407450199127, -0.011070162057876587, 0.04653465002775192, 0.05263819918036461, 0.04014687240123749, -0.011570101603865623, 0.0863010436296463, -0.0016188955632969737, -0.0972800999879837, 0.043253008276224136, 0.05158410593867302, 0.025566838...
1,960
1,960
['Balint Antal', 'Andras Hajdu']
1410.8577v1
Reliable microaneurysm detection in digital fundus images is still an open issue in medical image processing. We propose an ensemble-based framework to improve microaneurysm detection. Unlike the well-known approach of considering the output of multiple classifiers, we propose a combination of internal components of mi...
An Ensemble-based System for Microaneurysm Detection and Diabetic Retinopathy Grading
2,014
http://arxiv.org/pdf/1410.8577v1
Title Ensemblebased System Microaneurysm Detection Diabetic Retinopathy Grading Summary Reliable microaneurysm detection digital fundus image still open issue medical image processing propose ensemblebased framework improve microaneurysm detection Unlike wellknown approach considering output multiple classifier propose...
[0.018201958388090134, -0.02171334996819496, -0.007578928954899311, 0.008542180992662907, 0.0272004883736372, 0.03563505411148071, 0.054961901158094406, 0.04783674329519272, 0.03581336513161659, -0.01419894490391016, 0.08192003518342972, 0.01405976340174675, 0.020808303728699684, 0.04847802594304085, -0.000548334734048...
1,961
1,961
['Liang Lin', 'Guangrun Wang', 'Wangmeng Zuo', 'Xiangchu Feng', 'Lei Zhang']
1605.04039v1
Cross-domain visual data matching is one of the fundamental problems in many real-world vision tasks, e.g., matching persons across ID photos and surveillance videos. Conventional approaches to this problem usually involves two steps: i) projecting samples from different domains into a common space, and ii) computing (...
Cross-Domain Visual Matching via Generalized Similarity Measure and Feature Learning
2,016
http://arxiv.org/pdf/1605.04039v1
Title CrossDomain Visual Matching via Generalized Similarity Measure Feature Learning Summary Crossdomain visual data matching one fundamental problem many realworld vision task eg matching person across ID photo surveillance video Conventional approach problem usually involves two step projecting sample different doma...
[-0.03851548209786415, 0.057604432106018066, -0.004048462957143784, 0.01318285707384348, -0.025010228157043457, 0.023224834352731705, 0.047299742698669434, 0.00919927004724741, -0.028876138851046562, -0.003437710227444768, -0.052661728113889694, -0.060283757746219635, 0.024094179272651672, 0.03700126335024834, 0.034305...
1,962
1,962
['Chong Peng', 'Zhao Kang', 'Qiang Chen']
1609.08677v1
Robust principal component analysis (RPCA) has been widely used for recovering low-rank matrices in many data mining and machine learning problems. It separates a data matrix into a low-rank part and a sparse part. The convex approach has been well studied in the literature. However, state-of-the-art algorithms for the...
A Fast Factorization-based Approach to Robust PCA
2,016
http://arxiv.org/pdf/1609.08677v1
Title Fast Factorizationbased Approach Robust PCA Summary Robust principal component analysis RPCA widely used recovering lowrank matrix many data mining machine learning problem separate data matrix lowrank part sparse part convex approach well studied literature However stateoftheart algorithm convex approach usually...
[-0.00940763857215643, 0.044874463230371475, -0.039731215685606, 0.07160425931215286, 0.048297218978405, 0.0026042794343084097, -0.0028453918639570475, 0.019354255869984627, 0.0026609431952238083, -0.006012165453284979, 0.018118582665920258, 0.01902424916625023, 0.011185460723936558, 0.05133288353681564, -0.01697207055...
1,963
1,963
['Petar Veličković', 'Duo Wang', 'Nicholas D. Lane', 'Pietro Liò']
1610.00163v2
In this paper we propose cross-modal convolutional neural networks (X-CNNs), a novel biologically inspired type of CNN architectures, treating gradient descent-specialised CNNs as individual units of processing in a larger-scale network topology, while allowing for unconstrained information flow and/or weight sharing b...
X-CNN: Cross-modal Convolutional Neural Networks for Sparse Datasets
2,016
http://arxiv.org/pdf/1610.00163v2
Title XCNN Crossmodal Convolutional Neural Networks Sparse Datasets Summary paper propose crossmodal convolutional neural network XCNNs novel biologically inspired type CNN architecture treating gradient descentspecialised CNNs individual unit processing largerscale network topology allowing unconstrained information f...
[-0.007940697483718395, 0.02541801892220974, -0.011006438173353672, 0.06740560382604599, 0.004110641311854124, -0.04911426082253456, 0.05959277227520943, 0.012296239845454693, -0.0336618646979332, -0.0035540557000786066, -0.09771452844142914, 0.0011273830896243453, 0.0027520328294485807, 0.039823733270168304, 0.0477296...
1,964
1,964
['Rahaf Aljundi', 'Punarjay Chakravarty', 'Tinne Tuytelaars']
1611.06194v2
In this paper we introduce a model of lifelong learning, based on a Network of Experts. New tasks / experts are learned and added to the model sequentially, building on what was learned before. To ensure scalability of this process,data from previous tasks cannot be stored and hence is not available when learning a new...
Expert Gate: Lifelong Learning with a Network of Experts
2,016
http://arxiv.org/pdf/1611.06194v2
Title Expert Gate Lifelong Learning Network Experts Summary paper introduce model lifelong learning based Network Experts New task expert learned added model sequentially building learned ensure scalability processdata previous task cannot stored hence available learning new task critical issue context addressed litera...
[0.019863784313201904, 0.06267870217561722, -0.03421945124864578, 0.020915154367685318, -0.001748212962411344, -0.0005668086814694107, 0.06730546057224274, -0.013830186799168587, 0.00867999717593193, -0.05184248462319374, -0.015726320445537567, 0.026610907167196274, -0.0427088625729084, 0.1459474116563797, 0.0269659124...
1,965
1,965
['Ehsan Jahangiri', 'Erdem Yoruk', 'Rene Vidal', 'Laurent Younes', 'Donald Geman']
1701.02343v1
Despite enormous progress in object detection and classification, the problem of incorporating expected contextual relationships among object instances into modern recognition systems remains a key challenge. In this work we propose Information Pursuit, a Bayesian framework for scene parsing that combines prior models ...
Information Pursuit: A Bayesian Framework for Sequential Scene Parsing
2,017
http://arxiv.org/pdf/1701.02343v1
Title Information Pursuit Bayesian Framework Sequential Scene Parsing Summary Despite enormous progress object detection classification problem incorporating expected contextual relationship among object instance modern recognition system remains key challenge work propose Information Pursuit Bayesian framework scene p...
[0.02754591405391693, 0.0341215506196022, 0.06118111312389374, 0.09648868441581726, -0.050924960523843765, 0.062030695378780365, 0.007713215425610542, 0.04618518054485321, -0.03541554510593414, -0.06001687049865723, -0.009126164950430393, 0.016158491373062134, 0.020053774118423462, 0.015286733396351337, -0.034150194376...
1,966
1,966
['Ehsan Jahangiri', 'Alan L. Yuille']
1702.02258v2
We propose a method to generate multiple diverse and valid human pose hypotheses in 3D all consistent with the 2D detection of joints in a monocular RGB image. We use a novel generative model uniform (unbiased) in the space of anatomically plausible 3D poses. Our model is compositional (produces a pose by combining par...
Generating Multiple Diverse Hypotheses for Human 3D Pose Consistent with 2D Joint Detections
2,017
http://arxiv.org/pdf/1702.02258v2
Title Generating Multiple Diverse Hypotheses Human 3D Pose Consistent 2D Joint Detections Summary propose method generate multiple diverse valid human pose hypothesis 3D consistent 2D detection joint monocular RGB image use novel generative model uniform unbiased space anatomically plausible 3D pose model compositional...
[0.038782261312007904, 0.03645915910601616, -0.007229262497276068, 0.04087431728839874, -0.0036383606493473053, 0.028980283066630363, -0.00047003242070786655, 0.029281822964549065, 0.01399382296949625, -0.06658069789409637, -0.023100340738892555, -0.031910914927721024, 0.0664171576499939, 0.010348212905228138, 0.022117...
1,967
1,967
['Ahmed Hussain Qureshi', 'Yutaka Nakamura', 'Yuichiro Yoshikawa', 'Hiroshi Ishiguro']
1702.07492v1
For robots to coexist with humans in a social world like ours, it is crucial that they possess human-like social interaction skills. Programming a robot to possess such skills is a challenging task. In this paper, we propose a Multimodal Deep Q-Network (MDQN) to enable a robot to learn human-like interaction skills thr...
Robot gains Social Intelligence through Multimodal Deep Reinforcement Learning
2,017
http://arxiv.org/pdf/1702.07492v1
Title Robot gain Social Intelligence Multimodal Deep Reinforcement Learning Summary robot coexist human social world like crucial posse humanlike social interaction skill Programming robot posse skill challenging task paper propose Multimodal Deep QNetwork MDQN enable robot learn humanlike interaction skill trial error...
[0.02404920570552349, 0.03676782548427582, -0.005807043518871069, -0.0054890066385269165, 0.007047295104712248, 0.00698200473561883, 0.0013814045814797282, -0.044368576258420944, 0.0140475919470191, 0.003226795932278037, -0.060420677065849304, 0.03283054754137993, -0.0002511660859454423, 0.07016677409410477, 0.03045933...
1,968
1,968
['Ahmed Hussain Qureshi', 'Yutaka Nakamura', 'Yuichiro Yoshikawa', 'Hiroshi Ishiguro']
1702.08626v1
For a safe, natural and effective human-robot social interaction, it is essential to develop a system that allows a robot to demonstrate the perceivable responsive behaviors to complex human behaviors. We introduce the Multimodal Deep Attention Recurrent Q-Network using which the robot exhibits human-like social intera...
Show, Attend and Interact: Perceivable Human-Robot Social Interaction through Neural Attention Q-Network
2,017
http://arxiv.org/pdf/1702.08626v1
Title Show Attend Interact Perceivable HumanRobot Social Interaction Neural Attention QNetwork Summary safe natural effective humanrobot social interaction essential develop system allows robot demonstrate perceivable responsive behavior complex human behavior introduce Multimodal Deep Attention Recurrent QNetwork usin...
[0.036009062081575394, -0.010334296151995659, -0.0034927933011204004, -0.032354217022657394, 0.012577390298247337, 0.013225115835666656, 0.00267756637185812, -0.017654718831181526, 0.02671835757791996, -0.0011322794016450644, -0.021516337990760803, 0.01052994653582573, -0.01730543188750744, 0.0630308985710144, 0.027843...
1,969
1,969
['Jose Luis Garcia-Arroyo', 'Begonya Garcia-Zapirain']
1703.03888v1
This paper proposes an innovative method for segmentation of skin lesions in dermoscopy images developed by the authors, based on fuzzy classification of pixels and histogram thresholding.
Segmentation of skin lesions based on fuzzy classification of pixels and histogram thresholding
2,017
http://arxiv.org/pdf/1703.03888v1
Title Segmentation skin lesion based fuzzy classification pixel histogram thresholding Summary paper proposes innovative method segmentation skin lesion dermoscopy image developed author based fuzzy classification pixel histogram thresholding Authors 0 Ahmed Osman Wojciech Samek 1 Ji Young Lee Franck Dernoncourt 2 Iuli...
[0.010086758062243462, -0.030734796077013016, -0.02581441029906273, -0.022781111299991608, 0.0185412410646677, 0.0018030513310804963, 0.029919806867837906, 0.029927093535661697, 0.004885641857981682, 0.0009628386469557881, 0.0597672201693058, 0.015730326995253563, 0.02837780863046646, 0.006459304131567478, -0.008551144...
1,970
1,970
['Swami Sankaranarayanan', 'Arpit Jain', 'Ser Nam Lim']
1703.07928v2
Convolutional Neural Networks have been a subject of great importance over the past decade and great strides have been made in their utility for producing state of the art performance in many computer vision problems. However, the behavior of deep networks is yet to be fully understood and is still an active area of re...
Self corrective Perturbations for Semantic Segmentation and Classification
2,017
http://arxiv.org/pdf/1703.07928v2
Title Self corrective Perturbations Semantic Segmentation Classification Summary Convolutional Neural Networks subject great importance past decade great stride made utility producing state art performance many computer vision problem However behavior deep network yet fully understood still active area research work pr...
[0.014037585817277431, 0.027607379481196404, 0.0064954375848174095, 0.0861433893442154, -0.0035159692633897066, -0.009506097994744778, 0.008485223166644573, -0.03128369152545929, -0.031467217952013016, -0.007371558807790279, -0.03004472702741623, 0.0765342116355896, -0.031030364334583282, 0.01316814310848713, -0.003328...
1,971
1,971
['Hongyoon Choi', 'Kyong Hwan Jin']
1704.06033v1
For effective treatment of Alzheimer disease (AD), it is important to identify subjects who are most likely to exhibit rapid cognitive decline. Herein, we developed a novel framework based on a deep convolutional neural network which can predict future cognitive decline in mild cognitive impairment (MCI) patients using...
Predicting Cognitive Decline with Deep Learning of Brain Metabolism and Amyloid Imaging
2,017
http://arxiv.org/pdf/1704.06033v1
Title Predicting Cognitive Decline Deep Learning Brain Metabolism Amyloid Imaging Summary effective treatment Alzheimer disease AD important identify subject likely exhibit rapid cognitive decline Herein developed novel framework based deep convolutional neural network predict future cognitive decline mild cognitive im...
[-0.022972162812948227, 0.10184957087039948, -0.020820150151848793, -0.007596414070576429, 0.057178959250450134, 0.008113421499729156, 0.03202507272362709, 0.06400543451309204, 0.06622902303934097, 0.04947930946946144, 0.013972852379083633, 0.029944688081741333, 0.022608673200011253, 0.09221390634775162, 0.023791447281...
1,972
1,972
['Babak Toghiani-Rizi', 'Christofer Lind', 'Maria Svensson', 'Marcus Windmark']
1705.05884v1
In this report, an automated bartender system was developed for making orders in a bar using hand gestures. The gesture recognition of the system was developed using Machine Learning techniques, where the model was trained to classify gestures using collected data. The final model used in the system reached an average ...
Static Gesture Recognition using Leap Motion
2,017
http://arxiv.org/pdf/1705.05884v1
Title Static Gesture Recognition using Leap Motion Summary report automated bartender system developed making order bar using hand gesture gesture recognition system developed using Machine Learning technique model trained classify gesture using collected data final model used system reached average accuracy 95 system ...
[0.02472764626145363, 0.03799665346741676, -0.020114004611968994, -0.012377910315990448, -0.02267616055905819, 0.06293302774429321, 0.07161227613687515, 0.061751168221235275, 0.00851206574589014, -0.02691817842423916, 0.007224249187856913, -0.002497903537005186, 0.04215674102306366, 0.04655322805047035, -0.009609126485...
1,973
1,973
['Nikolay Jetchev', 'Urs Bergmann']
1709.04695v1
We present a novel method to solve image analogy problems : it allows to learn the relation between paired images present in training data, and then generalize and generate images that correspond to the relation, but were never seen in the training set. Therefore, we call the method Conditional Analogy Generative Adver...
The Conditional Analogy GAN: Swapping Fashion Articles on People Images
2,017
http://arxiv.org/pdf/1709.04695v1
Title Conditional Analogy GAN Swapping Fashion Articles People Images Summary present novel method solve image analogy problem allows learn relation paired image present training data generalize generate image correspond relation never seen training set Therefore call method Conditional Analogy Generative Adversarial N...
[0.012495940551161766, 0.07017572969198227, 0.0011922339908778667, 0.0726662129163742, -0.006574581377208233, 0.004304699599742889, 0.021316833794116974, 0.004403966479003429, -0.02447667345404625, -0.008010302670300007, -0.0257248654961586, 0.017762543633580208, -0.03588474914431572, 0.015225215815007687, 0.0513887628...
1,974
1,974
['Amit Mandelbaum', 'Daphna Weinshall']
1709.09844v1
The reliable measurement of confidence in classifiers' predictions is very important for many applications and is, therefore, an important part of classifier design. Yet, although deep learning has received tremendous attention in recent years, not much progress has been made in quantifying the prediction confidence of...
Distance-based Confidence Score for Neural Network Classifiers
2,017
http://arxiv.org/pdf/1709.09844v1
Title Distancebased Confidence Score Neural Network Classifiers Summary reliable measurement confidence classifier prediction important many application therefore important part classifier design Yet although deep learning received tremendous attention recent year much progress made quantifying prediction confidence ne...
[0.003964792937040329, 0.05895324423909187, -0.013620154932141304, 0.021847909316420555, -0.029156610369682312, -0.034118685871362686, 0.025903325527906418, -0.0137969134375453, -0.010856629349291325, 0.02672867476940155, 0.017470043152570724, -0.027551408857107162, 0.013855512253940105, 0.06485381722450256, 0.00253336...
1,975
1,975
['Han Zhang', 'Tao Xu', 'Hongsheng Li', 'Shaoting Zhang', 'Xiaogang Wang', 'Xiaolei Huang', 'Dimitris Metaxas']
1710.10916v2
Although Generative Adversarial Networks (GANs) have shown remarkable success in various tasks, they still face challenges in generating high quality images. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) aiming at generating high-resolution photo-realistic images. First, we propose a two-...
StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks
2,017
http://arxiv.org/pdf/1710.10916v2
Title StackGAN Realistic Image Synthesis Stacked Generative Adversarial Networks Summary Although Generative Adversarial Networks GANs shown remarkable success various task still face challenge generating high quality image paper propose Stacked Generative Adversarial Networks StackGAN aiming generating highresolution ...
[-0.0015631030546501279, 0.07733391225337982, 0.01014532521367073, 0.026093093678355217, 0.014250113628804684, -0.022589268162846565, 0.018427535891532898, 0.009084799326956272, -0.06361372023820877, 0.022160882130265236, -0.02895629219710827, -0.014986860565841198, -0.010389686562120914, 0.020656755194067955, 0.046363...
1,976
1,976
['Rahaf Aljundi', 'Francesca Babiloni', 'Mohamed Elhoseiny', 'Marcus Rohrbach', 'Tinne Tuytelaars']
1711.09601v1
Humans can learn in a continuous manner. Old rarely utilized knowledge can be overwritten by new incoming information while important, frequently used knowledge is prevented from being erased. In artificial learning systems, lifelong learning so far has focused mainly on accumulating knowledge over tasks and overcoming...
Memory Aware Synapses: Learning what (not) to forget
2,017
http://arxiv.org/pdf/1711.09601v1
Title Memory Aware Synapses Learning forget Summary Humans learn continuous manner Old rarely utilized knowledge overwritten new incoming information important frequently used knowledge prevented erased artificial learning system lifelong learning far focused mainly accumulating knowledge task overcoming catastrophic f...
[0.0028379769064486027, 0.034335386008024216, -0.020884960889816284, 0.03227798268198967, 0.020020270720124245, 0.012734457850456238, 0.014438924379646778, -0.02661140263080597, 0.0406244732439518, -0.011156192980706692, -0.008518180809915066, 0.0182108823210001, 0.0025903903879225254, 0.04846550151705742, 0.0282554700...
1,977
1,977
['Hyunwoo Lee', 'Jooyoung Kim', 'Dojun Yang', 'Joon-Ho Kim']
1711.11200v1
This paper proposes a real-time embedded fall detection system using a DVS(Dynamic Vision Sensor) that has never been used for traditional fall detection, a dataset for fall detection using that, and a DVS-TN(DVS-Temporal Network). The first contribution is building a DVS Falls Dataset, which made our network to recogn...
Embedded Real-Time Fall Detection Using Deep Learning For Elderly Care
2,017
http://arxiv.org/pdf/1711.11200v1
Title Embedded RealTime Fall Detection Using Deep Learning Elderly Care Summary paper proposes realtime embedded fall detection system using DVSDynamic Vision Sensor never used traditional fall detection dataset fall detection using DVSTNDVSTemporal Network first contribution building DVS Falls Dataset made network rec...
[-0.05720779299736023, -0.0075706010684370995, -0.030688539147377014, 0.039439376443624496, 0.011567014269530773, -0.010779167525470257, 0.005299276206642389, -0.007035915274173021, 0.03483397141098976, -0.005770990625023842, 0.0805526077747345, 0.01762794516980648, 0.013498914428055286, 0.08066699653863907, 0.01740673...
1,978
1,978
['Ruth Fong', 'Andrea Vedaldi']
1801.03454v1
In an effort to understand the meaning of the intermediate representations captured by deep networks, recent papers have tried to associate specific semantic concepts to individual neural network filter responses, where interesting correlations are often found, largely by focusing on extremal filter responses. In this ...
Net2Vec: Quantifying and Explaining how Concepts are Encoded by Filters in Deep Neural Networks
2,018
http://arxiv.org/pdf/1801.03454v1
Title Net2Vec Quantifying Explaining Concepts Encoded Filters Deep Neural Networks Summary effort understand meaning intermediate representation captured deep network recent paper tried associate specific semantic concept individual neural network filter response interesting correlation often found largely focusing ext...
[-0.005346860270947218, 0.041252121329307556, -0.0288819819688797, 0.043204862624406815, 0.0025437537115067244, 0.0035371461417526007, -0.019467273727059364, 0.015932800248265266, -0.06381386518478394, 0.0017182852607220411, 0.044278811663389206, 0.0178231094032526, 0.017481604591012, 0.08827931433916092, 0.02625123225...
1,979
1,979
['Deboleena Roy', 'Priyadarshini Panda', 'Kaushik Roy']
1802.05800v1
In recent years, Convolutional Neural Networks (CNNs) have shown remarkable performance in many computer vision tasks such as object recognition and detection. However, complex training issues, such as "catastrophic forgetting" and hyper-parameter tuning, make incremental learning in CNNs a difficult challenge. In this...
Tree-CNN: A Deep Convolutional Neural Network for Lifelong Learning
2,018
http://arxiv.org/pdf/1802.05800v1
Title TreeCNN Deep Convolutional Neural Network Lifelong Learning Summary recent year Convolutional Neural Networks CNNs shown remarkable performance many computer vision task object recognition detection However complex training issue catastrophic forgetting hyperparameter tuning make incremental learning CNNs difficu...
[0.023118233308196068, 0.09196504950523376, -0.0065264394506812096, 0.06721778213977814, 0.020465584471821785, 0.012314637191593647, -0.010618460364639759, 0.0011593463132157922, -0.06666744500398636, 0.013124784454703331, -0.0004200581170152873, 0.02488747425377369, -0.034330710768699646, 0.027935773134231567, -0.0297...
1,980
1,980
['Cecilia S. Lee', 'Ariel J. Tyring', 'Yue Wu', 'Sa Xiao', 'Ariel S. Rokem', 'Nicolaas P. Deruyter', 'Qinqin Zhang', 'Adnan Tufail', 'Ruikang K. Wang', 'Aaron Y. Lee']
1802.08925v1
Despite significant advances in artificial intelligence (AI) for computer vision, its application in medical imaging has been limited by the burden and limits of expert-generated labels. We used images from optical coherence tomography angiography (OCTA), a relatively new imaging modality that measures perfusion of the...
Generating retinal flow maps from structural optical coherence tomography with artificial intelligence
2,018
http://arxiv.org/pdf/1802.08925v1
Title Generating retinal flow map structural optical coherence tomography artificial intelligence Summary Despite significant advance artificial intelligence AI computer vision application medical imaging limited burden limit expertgenerated label used image optical coherence tomography angiography OCTA relatively new ...
[-0.015590861439704895, -0.008160259574651718, -0.0069184978492558, -0.03464813157916069, -0.03814835473895073, -0.014395536854863167, 0.03322206065058708, 0.016522180289030075, -0.0010247487807646394, 0.027023112401366234, 0.06176641955971718, 0.017349539324641228, 0.02651374414563179, 0.1433231085538864, -0.016031323...
1,981
1,981
['Sathya N. Ravi', 'Ronak Mehta', 'Vikas Singh']
1803.08137v1
We revisit the Blind Deconvolution problem with a focus on understanding its robustness and convergence properties. Provable robustness to noise and other perturbations is receiving recent interest in vision, from obtaining immunity to adversarial attacks to assessing and describing failure modes of algorithms in missi...
Robust Blind Deconvolution via Mirror Descent
2,018
http://arxiv.org/pdf/1803.08137v1
Title Robust Blind Deconvolution via Mirror Descent Summary revisit Blind Deconvolution problem focus understanding robustness convergence property Provable robustness noise perturbation receiving recent interest vision obtaining immunity adversarial attack assessing describing failure mode algorithm mission critical a...
[-0.024873217567801476, 0.07485316693782806, -0.005499345250427723, 0.059864338487386703, 0.015645965933799744, -0.029218554496765137, 0.033223606646060944, 0.006737859919667244, 0.023283330723643303, 0.030895516276359558, -0.01648012362420559, 0.03235667198896408, 0.020969627425074577, 0.059272076934576035, 0.01201226...
1,982
1,982
['Roland Memisevic']
1110.0107v2
A fundamental operation in many vision tasks, including motion understanding, stereopsis, visual odometry, or invariant recognition, is establishing correspondences between images or between images and data from other modalities. We present an analysis of the role that multiplicative interactions play in learning such ...
Learning to relate images: Mapping units, complex cells and simultaneous eigenspaces
2,011
http://arxiv.org/pdf/1110.0107v2
Title Learning relate image Mapping unit complex cell simultaneous eigenspaces Summary fundamental operation many vision task including motion understanding stereopsis visual odometry invariant recognition establishing correspondence image image data modality present analysis role multiplicative interaction play learni...
[-0.004944073501974344, 0.018909916281700134, -0.0016458469908684492, 0.027415376156568527, -0.00268951547332108, 0.06175542622804642, 0.040185898542404175, 0.01638473942875862, -0.03559820353984833, -0.017726033926010132, -0.021726436913013458, -0.011056032963097095, 0.059657130390405655, 0.00103889056481421, 0.012414...
1,983
1,983
['Shehroze Bhatti', 'Alban Desmaison', 'Ondrej Miksik', 'Nantas Nardelli', 'N. Siddharth', 'Philip H. S. Torr']
1612.00380v1
A number of recent approaches to policy learning in 2D game domains have been successful going directly from raw input images to actions. However when employed in complex 3D environments, they typically suffer from challenges related to partial observability, combinatorial exploration spaces, path planning, and a scarc...
Playing Doom with SLAM-Augmented Deep Reinforcement Learning
2,016
http://arxiv.org/pdf/1612.00380v1
Title Playing Doom SLAMAugmented Deep Reinforcement Learning Summary number recent approach policy learning 2D game domain successful going directly raw input image action However employed complex 3D environment typically suffer challenge related partial observability combinatorial exploration space path planning scarc...
[0.03214658424258232, 0.005788092967122793, 0.0007911993889138103, -0.00891051534563303, 0.018258610740303993, -0.012062234804034233, -0.02829083800315857, -0.056368015706539154, -0.022725114598870277, 0.006848024670034647, -0.05227147787809372, 0.03324265405535698, -0.0011649943189695477, 0.08107288926839828, 0.027555...
1,984
1,984
['Parker Koch', 'Jason J. Corso']
1612.04468v1
Whereas CNNs have demonstrated immense progress in many vision problems, they suffer from a dependence on monumental amounts of labeled training data. On the other hand, dictionary learning does not scale to the size of problems that CNNs can handle, despite being very effective at low-level vision tasks such as denois...
Sparse Factorization Layers for Neural Networks with Limited Supervision
2,016
http://arxiv.org/pdf/1612.04468v1
Title Sparse Factorization Layers Neural Networks Limited Supervision Summary Whereas CNNs demonstrated immense progress many vision problem suffer dependence monumental amount labeled training data hand dictionary learning scale size problem CNNs handle despite effective lowlevel vision task denoising inpainting Recen...
[0.00541137158870697, 0.046095382422208786, 0.02151351422071457, 0.06641347706317902, 0.031929466873407364, -0.007964573800563812, 0.04462100565433502, 0.024983523413538933, -0.05556724593043327, 0.01293788943439722, -0.02247467078268528, 0.003511303337290883, -0.018398568034172058, 0.060777559876441956, 0.000833452155...
1,985
1,985
['Alberto N. Escalante-B.', 'Laurenz Wiskott']
1509.08329v1
Slow feature analysis (SFA) is an unsupervised learning algorithm that extracts slowly varying features from a time series. Graph-based SFA (GSFA) is a supervised extension that can solve regression problems if followed by a post-processing regression algorithm. A training graph specifies arbitrary connections between ...
Theoretical Analysis of the Optimal Free Responses of Graph-Based SFA for the Design of Training Graphs
2,015
http://arxiv.org/pdf/1509.08329v1
Title Theoretical Analysis Optimal Free Responses GraphBased SFA Design Training Graphs Summary Slow feature analysis SFA unsupervised learning algorithm extract slowly varying feature time series Graphbased SFA GSFA supervised extension solve regression problem followed postprocessing regression algorithm training gra...
[0.023135729134082794, 0.03272026777267456, -0.02626785822212696, 0.029738018289208412, 0.010589911602437496, 0.010526339523494244, 0.06106388568878174, 0.015787379816174507, 0.058025915175676346, -0.006170670501887798, 0.03669169917702675, 0.008816413581371307, 0.03022926114499569, 0.12708470225334167, 0.0152737349271...
1,986
1,986
['Agne Grinciunaite', 'Amogh Gudi', 'Emrah Tasli', 'Marten den Uyl']
1609.00036v3
This paper explores the capabilities of convolutional neural networks to deal with a task that is easily manageable for humans: perceiving 3D pose of a human body from varying angles. However, in our approach, we are restricted to using a monocular vision system. For this purpose, we apply a convolutional neural networ...
Human Pose Estimation in Space and Time using 3D CNN
2,016
http://arxiv.org/pdf/1609.00036v3
Title Human Pose Estimation Space Time using 3D CNN Summary paper explores capability convolutional neural network deal task easily manageable human perceiving 3D pose human body varying angle However approach restricted using monocular vision system purpose apply convolutional neural network approach RGB video extend ...
[0.026248063892126083, 0.029549134895205498, 0.007911465130746365, 0.04756024852395058, 0.005172345787286758, 0.015959586948156357, 0.022854674607515335, 0.003499396611005068, -0.01535397581756115, -0.033727191388607025, -0.011850519105792046, -0.03264998644590378, 0.042170342057943344, 0.03846743330359459, 0.080222636...
1,987
1,987
['Amal Rannen Triki', 'Rahaf Aljundi', 'Mathew B. Blaschko', 'Tinne Tuytelaars']
1704.01920v1
This paper introduces a new lifelong learning solution where a single model is trained for a sequence of tasks. The main challenge that vision systems face in this context is catastrophic forgetting: as they tend to adapt to the most recently seen task, they lose performance on the tasks that were learned previously. O...
Encoder Based Lifelong Learning
2,017
http://arxiv.org/pdf/1704.01920v1
Title Encoder Based Lifelong Learning Summary paper introduces new lifelong learning solution single model trained sequence task main challenge vision system face context catastrophic forgetting tend adapt recently seen task lose performance task learned previously method aim preserving knowledge previous task learning...
[-0.02078273333609104, 0.07169186323881149, -0.010402734391391277, 0.024157391861081123, -0.0023549431934952736, 0.033892564475536346, 0.03565932810306549, -0.00938281137496233, -0.006047660484910011, -0.0030452036298811436, 0.04142463952302933, 0.04197240248322487, -0.004235920961946249, 0.10904201865196228, 0.0030975...
1,988
1,988
['Seyed A Sajjadi', 'Danial Moazen', 'Ani Nahapetian']
1705.02689v1
Wearable computing is one of the fastest growing technologies today. Smart watches are poised to take over at least of half the wearable devices market in the near future. Smart watch screen size, however, is a limiting factor for growth, as it restricts practical text input. On the other hand, wearable devices have so...
AirDraw: Leveraging Smart Watch Motion Sensors for Mobile Human Computer Interactions
2,017
http://arxiv.org/pdf/1705.02689v1
Title AirDraw Leveraging Smart Watch Motion Sensors Mobile Human Computer Interactions Summary Wearable computing one fastest growing technology today Smart watch poised take least half wearable device market near future Smart watch screen size however limiting factor growth restricts practical text input hand wearable...
[-0.007691414561122656, -0.05274724215269089, -0.007116193417459726, 0.039587561041116714, -0.00291139492765069, 0.03348410502076149, 0.017813069745898247, 0.04886007681488991, 0.036154769361019135, -0.04486796259880066, 0.034991879016160965, 0.045356422662734985, 0.0635870173573494, 0.060966283082962036, 0.00176254822...
1,989
1,989
['David Rolnick', 'Yaron Meirovitch', 'Toufiq Parag', 'Hanspeter Pfister', 'Viren Jain', 'Jeff W. Lichtman', 'Edward S. Boyden', 'Nir Shavit']
1705.10882v1
Deep learning algorithms for connectomics rely upon localized classification, rather than overall morphology. This leads to a high incidence of erroneously merged objects. Humans, by contrast, can easily detect such errors by acquiring intuition for the correct morphology of objects. Biological neurons have complicated...
Morphological Error Detection in 3D Segmentations
2,017
http://arxiv.org/pdf/1705.10882v1
Title Morphological Error Detection 3D Segmentations Summary Deep learning algorithm connectomics rely upon localized classification rather overall morphology lead high incidence erroneously merged object Humans contrast easily detect error acquiring intuition correct morphology object Biological neuron complicated var...
[-0.01790752448141575, 0.020751740783452988, -0.009770575910806656, 0.037233684211969376, 0.027018265798687935, 0.0245415810495615, 0.02749733254313469, -0.02139386348426342, -0.023958662524819374, 0.054385874420404434, 0.02465571090579033, -0.030502479523420334, 0.09292641282081604, 0.047948263585567474, 0.03407514467...
1,990
1,990
['Li Shen']
1708.09427v3
We develop an end-to-end training algorithm for whole-image breast cancer diagnosis based on mammograms. It requires lesion annotations only at the first stage of training. After that, a whole image classifier can be trained using only image level labels. This greatly reduced the reliance on lesion annotations. Our app...
End-to-end Training for Whole Image Breast Cancer Diagnosis using An All Convolutional Design
2,017
http://arxiv.org/pdf/1708.09427v3
Title Endtoend Training Whole Image Breast Cancer Diagnosis using Convolutional Design Summary develop endtoend training algorithm wholeimage breast cancer diagnosis based mammogram requires lesion annotation first stage training whole image classifier trained using image level label greatly reduced reliance lesion ann...
[0.03249354660511017, 0.03197140619158745, -0.009393274784088135, 0.011779106222093105, 0.019147001206874847, 0.032474350184202194, 0.034797534346580505, 0.031608983874320984, -0.007411726750433445, 0.019252043217420578, 0.02239890955388546, -0.01690598390996456, -0.053833700716495514, 0.030426815152168274, -0.01224809...
1,991
1,991
['Vijay Manikandan Janakiraman']
1710.04749v2
Although aviation accidents are rare, safety incidents occur more frequently and require a careful analysis to detect and mitigate risks in a timely manner. Analyzing safety incidents using operational data and producing event-based explanations is invaluable to airline companies as well as to governing organizations s...
Explaining Aviation Safety Incidents Using Deep Temporal Multiple Instance Learning
2,017
http://arxiv.org/pdf/1710.04749v2
Title Explaining Aviation Safety Incidents Using Deep Temporal Multiple Instance Learning Summary Although aviation accident rare safety incident occur frequently require careful analysis detect mitigate risk timely manner Analyzing safety incident using operational data producing eventbased explanation invaluable airl...
[-0.03841226175427437, 0.032490313053131104, -0.0185301024466753, 0.03434028849005699, 0.013614344410598278, 0.03312031924724579, 0.02941741794347763, -0.011409520171582699, -0.06090288609266281, -0.00532841170206666, 0.06072196736931801, -0.01810302957892418, -0.0344916433095932, 0.09089121967554092, 0.018178563565015...
1,992
1,992
['Housam Khalifa Bashier Babiker', 'Randy Goebel']
1711.06431v2
We present a method for explaining the image classification predictions of deep convolution neural networks, by highlighting the pixels in the image which influence the final class prediction. Our method requires the identification of a heuristic method to select parameters hypothesized to be most relevant in this pred...
Using KL-divergence to focus Deep Visual Explanation
2,017
http://arxiv.org/pdf/1711.06431v2
Title Using KLdivergence focus Deep Visual Explanation Summary present method explaining image classification prediction deep convolution neural network highlighting pixel image influence final class prediction method requires identification heuristic method select parameter hypothesized relevant prediction use Kullbac...
[0.007835954427719116, 0.00448676198720932, -0.02128908969461918, 0.04125114530324936, -0.00716746598482132, -0.023615358397364616, 0.019926849752664566, 0.007102197501808405, -0.06962113082408905, 0.006123377475887537, 0.013773586601018906, 0.03955304995179176, 0.007022903300821781, 0.043635331094264984, 0.04294789955...
1,993
1,993
['Xuelin Qian', 'Yanwei Fu', 'Wenxuan Wang', 'Tao Xiang', 'Yang Wu', 'Yu-Gang Jiang', 'Xiangyang Xue']
1712.02225v5
Person Re-identification (re-id) faces two major challenges: the lack of cross-view paired training data and learning discriminative identity-sensitive and view-invariant features in the presence of large pose variations. In this work, we address both problems by proposing a novel deep person image generation model for...
Pose-Normalized Image Generation for Person Re-identification
2,017
http://arxiv.org/pdf/1712.02225v5
Title PoseNormalized Image Generation Person Reidentification Summary Person Reidentification reid face two major challenge lack crossview paired training data learning discriminative identitysensitive viewinvariant feature presence large pose variation work address problem proposing novel deep person image generation ...
[-0.02247706800699234, 0.09542440623044968, -0.011474188417196274, 0.0008663114276714623, 0.014104396104812622, 0.03762594237923622, 0.02607634849846363, 0.0017286693910136819, -0.00544303935021162, 0.0431060865521431, 0.007860760204494, -0.033278223127126694, -0.01388423703610897, -0.0014908039011061192, 0.06006785482...
1,994
1,994
['Mehdi S. M. Sajjadi', 'Raviteja Vemulapalli', 'Matthew Brown']
1801.04590v3
Recent advances in video super-resolution have shown that convolutional neural networks combined with motion compensation are able to merge information from multiple low-resolution (LR) frames to generate high-quality images. Current state-of-the-art methods process a batch of LR frames to generate a single high-resolu...
Frame-Recurrent Video Super-Resolution
2,018
http://arxiv.org/pdf/1801.04590v3
Title FrameRecurrent Video SuperResolution Summary Recent advance video superresolution shown convolutional neural network combined motion compensation able merge information multiple lowresolution LR frame generate highquality image Current stateoftheart method process batch LR frame generate single highresolution HR ...
[-0.019336925819516182, 0.02537389285862446, 0.049933332949876785, 0.02165704406797886, 0.01755663938820362, -0.00939319096505642, 0.015031899325549603, 0.059534259140491486, -0.06384757161140442, -0.009837331250309944, 0.04688708856701851, 0.00695439800620079, 0.00702296057716012, -0.008783289231359959, 0.034754484891...
1,995
1,995
['Alexey Chaplygin', 'Joshua Chacksfield']
1802.01435v1
Convolutional Neural Networks are a well-known staple of modern image classification. However, it can be difficult to assess the quality and robustness of such models. Deep models are known to perform well on a given training and estimation set, but can easily be fooled by data that is specifically generated for the pu...
A Method for Restoring the Training Set Distribution in an Image Classifier
2,018
http://arxiv.org/pdf/1802.01435v1
Title Method Restoring Training Set Distribution Image Classifier Summary Convolutional Neural Networks wellknown staple modern image classification However difficult ass quality robustness model Deep model known perform well given training estimation set easily fooled data specifically generated purpose shown one prod...
[-0.02852965146303177, 0.06851203739643097, -0.025401484221220016, 0.04940428584814072, -0.015637975186109543, -0.014230511151254177, 0.02036856859922409, 0.032664474099874496, -0.06536754965782166, 0.0053740390576422215, 0.020694907754659653, 0.044342491775751114, 0.00504992576315999, 0.041147295385599136, 0.027416577...
1,996
1,996
['Richard J. Preen', 'Larry Bull']
1201.5604v2
A number of representation schemes have been presented for use within learning classifier systems, ranging from binary encodings to neural networks. This paper presents results from an investigation into using discrete and fuzzy dynamical system representations within the XCSF learning classifier system. In particular,...
Discrete and fuzzy dynamical genetic programming in the XCSF learning classifier system
2,012
http://arxiv.org/pdf/1201.5604v2
Title Discrete fuzzy dynamical genetic programming XCSF learning classifier system Summary number representation scheme presented use within learning classifier system ranging binary encoding neural network paper present result investigation using discrete fuzzy dynamical system representation within XCSF learning clas...
[-0.0336943082511425, 0.03730117157101631, -0.06468050926923752, -0.02906373329460621, -0.007885068655014038, -0.02608029730618, -0.009982495568692684, 0.026202762499451637, -0.013336293399333954, -0.01947295106947422, 0.08659270405769348, 0.011125048622488976, -0.015593606047332287, 0.08251720666885376, -0.02254223264...
1,997
1,997
['Zhen Li', 'Yizhou Yu']
1604.07176v1
Protein secondary structure prediction is an important problem in bioinformatics. Inspired by the recent successes of deep neural networks, in this paper, we propose an end-to-end deep network that predicts protein secondary structures from integrated local and global contextual features. Our deep architecture leverage...
Protein Secondary Structure Prediction Using Cascaded Convolutional and Recurrent Neural Networks
2,016
http://arxiv.org/pdf/1604.07176v1
Title Protein Secondary Structure Prediction Using Cascaded Convolutional Recurrent Neural Networks Summary Protein secondary structure prediction important problem bioinformatics Inspired recent success deep neural network paper propose endtoend deep network predicts protein secondary structure integrated local global...
[0.02827388234436512, 0.029116561636328697, 0.011595779098570347, 0.03169550374150276, 0.023811565712094307, -0.025512278079986572, -0.008289695717394352, -0.009067535400390625, 0.0033934114035218954, 0.01916162669658661, 0.016912853345274925, -0.04536593705415726, 0.03575683757662773, 0.06950857490301132, 0.0296377837...
1,998
1,998
['Daniel Crawford', 'Anna Levit', 'Navid Ghadermarzy', 'Jaspreet S. Oberoi', 'Pooya Ronagh']
1612.05695v2
We investigate whether quantum annealers with select chip layouts can outperform classical computers in reinforcement learning tasks. We associate a transverse field Ising spin Hamiltonian with a layout of qubits similar to that of a deep Boltzmann machine (DBM) and use simulated quantum annealing (SQA) to numerically ...
Reinforcement Learning Using Quantum Boltzmann Machines
2,016
http://arxiv.org/pdf/1612.05695v2
Title Reinforcement Learning Using Quantum Boltzmann Machines Summary investigate whether quantum annealers select chip layout outperform classical computer reinforcement learning task associate transverse field Ising spin Hamiltonian layout qubits similar deep Boltzmann machine DBM use simulated quantum annealing SQA ...
[-0.03310352563858032, -0.017029354348778725, -0.056926388293504715, 0.025781957432627678, -0.03580565005540848, -0.015372546389698982, 8.114010415738449e-05, 0.017400646582245827, -0.0033552302047610283, 0.02108965814113617, -0.018162701278924942, 0.01810094155371189, -0.0388457216322422, 0.02690531313419342, 0.025603...
1,999
1,999
['Keki M. Burjorjee']
1307.3824v1
This paper establishes theoretical bonafides for implicit concurrent multivariate effect evaluation--implicit concurrency for short---a broad and versatile computational learning efficiency thought to underlie general-purpose, non-local, noise-tolerant optimization in genetic algorithms with uniform crossover (UGAs). W...
The Fundamental Learning Problem that Genetic Algorithms with Uniform Crossover Solve Efficiently and Repeatedly As Evolution Proceeds
2,013
http://arxiv.org/pdf/1307.3824v1
Title Fundamental Learning Problem Genetic Algorithms Uniform Crossover Solve Efficiently Repeatedly Evolution Proceeds Summary paper establishes theoretical bonafides implicit concurrent multivariate effect evaluationimplicit concurrency shorta broad versatile computational learning efficiency thought underlie general...
[-0.03447984531521797, 0.060624007135629654, -0.029828805476427078, -0.0012971253599971533, -0.0243133082985878, 0.011968517675995827, 0.03353654965758324, 0.011371093802154064, 0.00044509785948321223, -0.035329144448041916, 0.015419398434460163, 0.05531609430909157, -0.01644636131823063, 0.043388281017541885, -0.02370...