title
stringlengths
15
81
paper_url
stringlengths
31
31
authors
listlengths
1
6
type
stringclasses
2 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
616
1.37k
keywords
listlengths
0
0
TL;DR
large_stringclasses
0 values
submission_number
int64
1
42
arxiv_id
stringlengths
9
9
arxiv_id_source
stringclasses
1 value
embedding
listlengths
768
768
Word Representations via Gaussian Embedding
https://arxiv.org/abs/1412.6623
[ "Luke Vilnis", "Andrew McCallum" ]
Oral
null
Current work in lexical distributed representations maps each word to a point vector in low-dimensional space. Mapping instead to a density provides many interesting advantages, including better capturing uncertainty about a representation and its relationships, expressing asymmetries more naturally than dot product ...
[]
null
1
1412.6623
iclr_archive
[ -0.022525973618030548, -0.007958680391311646, 0.0029189216438680887, 0.06130314990878105, 0.026897192001342773, 0.05592765286564827, 0.02717588283121586, 0.0026656747795641422, -0.002198993694037199, -0.03068809024989605, -0.009585809893906116, 0.015337381511926651, -0.07212560623884201, 0...
Deep Captioning with Multimodal Recurrent Neural Networks (m-RNN)
https://arxiv.org/abs/1412.6632
[ "Junhua Mao", "Wei Xu", "Yi Yang", "Jiang Wang", "Alan Yuille" ]
Oral
null
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-netw...
[]
null
2
1412.6632
iclr_archive
[ -0.009589101187884808, -0.041977301239967346, -0.015281761065125465, 0.07077564299106598, 0.03211687132716179, 0.03752228990197182, 0.00904516689479351, 0.0399647057056427, -0.05049708113074303, -0.01798888109624386, -0.03508518263697624, 0.029002578929066658, -0.05790695920586586, -0.0007...
Deep Structured Output Learning for Unconstrained Text Recognition
https://arxiv.org/abs/1412.5903
[ "Max Jaderberg", "Karen Simonyan", "Andrea Vedaldi", "Andrew Zisserman" ]
Oral
null
We develop a representation suitable for the unconstrained recognition of words in natural images: the general case of no fixed lexicon and unknown length. To this end we propose a convolutional neural network (CNN) based architecture which incorporates a Conditional Random Field (CRF) graphical model, taking the w...
[]
null
3
1412.5903
iclr_archive
[ -0.00413596211001277, -0.027735279873013496, -0.004071063827723265, 0.0515449196100235, 0.033686406910419464, 0.029282517731189728, -0.0011436325730755925, 0.03766223043203354, 0.0021310923621058464, -0.02687574177980423, -0.027515815570950508, 0.033219192177057266, -0.0742066279053688, -0...
Very Deep Convolutional Networks for Large-Scale Image Recognition
https://arxiv.org/abs/1409.1556
[ "Karen Simonyan", "Andrew Zisserman" ]
Oral
null
In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improve...
[]
null
4
1409.1556
iclr_archive
[ 0.01817971095442772, -0.0473996140062809, 0.004819185007363558, 0.02936873771250248, 0.03162720799446106, 0.015696119517087936, 0.008704915642738342, 0.022175777703523636, -0.01362753938883543, -0.054987773299217224, 0.0107796099036932, -0.020215651020407677, -0.07133769243955612, 0.024413...
Fast Convolutional Nets With fbfft: A GPU Performance Evaluation
https://arxiv.org/abs/1412.7580
[ "Nicolas Vasilache", "Jeff Johnson", "Michael Mathieu", "Soumith Chintala", "Serkan Piantino", "Yann LeCun" ]
Oral
null
We examine the performance profile of Convolutional Neural Network training on the current generation of NVIDIA Graphics Processing Units. We introduce two new Fast Fourier Transform convolution implementations: one based on NVIDIA's cuFFT library, and another based on a Facebook authored FFT implementation, fbfft, t...
[]
null
5
1412.7580
iclr_archive
[ 0.005991718731820583, -0.04726449027657509, 0.025601275265216827, 0.025315532460808754, 0.03517220914363861, 0.03579822927713394, -0.012910126708447933, 0.049497686326503754, -0.007204278372228146, -0.05632399022579193, 0.015573403798043728, 0.0014487336156889796, -0.07538507133722305, -0....
Reweighted Wake-Sleep
https://arxiv.org/abs/1406.2751
[ "Jorg Bornschein", "Yoshua Bengio" ]
Oral
null
Training deep directed graphical models with many hidden variables and performing inference remains a major challenge. Helmholtz machines and deep belief networks are such models, and the wake-sleep algorithm has been proposed to train them. The wake-sleep algorithm relies on training not just the directed generative...
[]
null
6
1406.2751
iclr_archive
[ -0.006134308874607086, -0.0039626797661185265, -0.010186558589339256, 0.031610701233148575, 0.028113123029470444, 0.01993045024573803, 0.04788599908351898, 0.012760824523866177, 0.00452772993594408, -0.04028919339179993, -0.008980482816696167, 0.017194831743836403, -0.06251934915781021, -0...
The local low-dimensionality of natural images
https://arxiv.org/abs/1412.6626
[ "Olivier Henaff", "Johannes Balle", "Neil Rabinowitz", "Eero Simoncelli" ]
Oral
null
We develop a new statistical model for photographic images, in which the local responses of a bank of linear filters are described as jointly Gaussian, with zero mean and a covariance that varies slowly over spatial position. We optimize sets of filters so as to minimize the nuclear norms of matrices of their local a...
[]
null
7
1412.6626
iclr_archive
[ -0.0010592426406219602, -0.0028493625577539206, 0.016646448522806168, 0.0419737882912159, 0.0419563390314579, 0.04619161784648895, 0.008415493182837963, -0.019615905359387398, -0.06176786497235298, -0.07221227884292603, -0.006884079892188311, -0.010244947858154774, -0.0753149539232254, -0....
Memory Networks
https://arxiv.org/abs/1410.3916
[ "Jason Weston", "Sumit Chopra", "Antoine Bordes" ]
Oral
null
We describe a new class of learning models called memory networks. Memory networks reason with inference components combined with a long-term memory component; they learn how to use these jointly. The long-term memory can be read and written to, with the goal of using it for prediction. We investigate these models in...
[]
null
8
1410.3916
iclr_archive
[ -0.009670330211520195, 0.001341706607490778, -0.009222445078194141, 0.0511188805103302, 0.04864468052983284, 0.01231493428349495, -0.005306434817612171, 0.021064434200525284, -0.0411897748708725, 0.010568782687187195, 0.0015729879960417747, 0.024641918018460274, -0.04645440727472305, -0.01...
Object detectors emerge in Deep Scene CNNs
https://arxiv.org/abs/1412.6856
[ "Bolei Zhou", "Aditya Khosla", "Agata Lapedriza", "Aude Oliva", "Antonio Torralba" ]
Oral
null
With the success of new computational architectures for visual processing, such as convolutional neural networks (CNN) and access to image databases with millions of labeled examples (e.g., ImageNet, Places), the state of the art in computer vision is advancing rapidly. One important factor for continued progress is ...
[]
null
9
1412.6856
iclr_archive
[ -0.006357009056955576, 0.001259227399714291, 0.02100524492561817, 0.04599931463599205, 0.024247700348496437, 0.010110636241734028, 0.010036276653409004, 0.01003933697938919, -0.04300607740879059, -0.03792363777756691, -0.03891061991453171, 0.006014237646013498, -0.05411538854241371, -0.002...
Qualitatively characterizing neural network optimization problems
https://arxiv.org/abs/1412.6544
[ "Ian Goodfellow", "Oriol Vinyals" ]
Oral
null
Training neural networks involves solving large-scale non-convex optimization problems. This task has long been believed to be extremely difficult, with fear of local minima and other obstacles motivating a variety of schemes to improve optimization, such as unsupervised pretraining. However, modern neural networks a...
[]
null
10
1412.6544
iclr_archive
[ -0.04026986286044121, -0.022642264142632484, -0.005983470473438501, 0.04811333492398262, 0.03697055205702782, 0.050341468304395676, -0.0063139148987829685, -0.016202392056584358, -0.03586403653025627, -0.03219089284539223, -0.019967157393693924, 0.022246871143579483, -0.0397377647459507, 0...
Neural Machine Translation by Jointly Learning to Align and Translate
https://arxiv.org/abs/1409.0473
[ "Dzmitry Bahdanau", "Kyunghyun Cho", "Yoshua Bengio" ]
Oral
null
Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural ...
[]
null
11
1409.0473
iclr_archive
[ -0.005684528965502977, -0.03467094525694847, -0.007944873534142971, 0.018774930387735367, 0.016290921717882156, 0.06396574527025223, 0.015409497544169426, 0.02249527908861637, -0.01750517264008522, -0.04119409993290901, -0.031601957976818085, 0.03298115357756615, -0.03783518448472023, 0.00...
FitNets: Hints for Thin Deep Nets
https://arxiv.org/abs/1412.6550
[ "Adriana Romero", "Nicolas Ballas", "Samira Ebrahimi Kahou", "Antoine Chassang", "Carlo Gatta", "Yoshua Bengio" ]
Poster
null
While depth tends to improve network performances, it also makes gradient-based training more difficult since deeper networks tend to be more non-linear. The recently proposed knowledge distillation approach is aimed at obtaining small and fast-to-execute models, and it has shown that a student network could imitate ...
[]
null
12
1412.6550
iclr_archive
[ 0.02412000112235546, -0.05426397547125816, -0.01909446343779564, 0.04840005934238434, 0.08811463415622711, 0.008961433544754982, 0.020987311378121376, -0.022420553490519524, -0.007944815792143345, -0.024905845522880554, -0.009533978998661041, -0.0008167215855792165, -0.05341830849647522, 0...
Techniques for Learning Binary Stochastic Feedforward Neural Networks
https://arxiv.org/abs/1406.2989
[ "Tapani Raiko", "Mathias Berglund", "Guillaume Alain", "Laurent Dinh" ]
Poster
null
Stochastic binary hidden units in a multi-layer perceptron (MLP) network give at least three potential benefits when compared to deterministic MLP networks. (1) They allow to learn one-to-many type of mappings. (2) They can be used in structured prediction problems, where modeling the internal structure of the output...
[]
null
13
1406.2989
iclr_archive
[ -0.01394066121429205, -0.01628882624208927, -0.015613674186170101, 0.045353859663009644, 0.017043476924300194, 0.04817509278655052, 0.01839597336947918, 0.02773919329047203, -0.026729939505457878, -0.03437076881527901, 0.005275174044072628, -0.009025489911437035, -0.04710961878299713, -0.0...
Reweighted Wake-Sleep
https://arxiv.org/abs/1406.2751
[ "Jorg Bornschein", "Yoshua Bengio" ]
Poster
null
Training deep directed graphical models with many hidden variables and performing inference remains a major challenge. Helmholtz machines and deep belief networks are such models, and the wake-sleep algorithm has been proposed to train them. The wake-sleep algorithm relies on training not just the directed generative...
[]
null
14
1406.2751
iclr_archive
[ -0.006134308874607086, -0.0039626797661185265, -0.010186558589339256, 0.031610701233148575, 0.028113123029470444, 0.01993045024573803, 0.04788599908351898, 0.012760824523866177, 0.00452772993594408, -0.04028919339179993, -0.008980482816696167, 0.017194831743836403, -0.06251934915781021, -0...
Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs
https://arxiv.org/abs/1412.7062
[ "Liang-Chieh Chen", "George Papandreou", "Iasonas Kokkinos", "Kevin Murphy", "Alan Yuille" ]
Poster
null
Deep Convolutional Neural Networks (DCNNs) have recently shown state of the art performance in high level vision tasks, such as image classification and object detection. This work brings together methods from DCNNs and probabilistic graphical models for addressing the task of pixel-level classification (also called ...
[]
null
15
1412.7062
iclr_archive
[ -0.011655132286250591, -0.03302793204784393, 0.007391819264739752, 0.05146004259586334, 0.022414250299334526, 0.033468760550022125, 0.0029790091793984175, 0.03238249942660332, -0.03225111588835716, -0.04284127429127693, -0.04488544166088104, -0.011368717066943645, -0.03390265256166458, 0.0...
Multiple Object Recognition with Visual Attention
https://arxiv.org/abs/1412.7755
[ "Jimmy Ba", "Volodymyr Mnih", "Koray Kavukcuoglu" ]
Poster
null
We present an attention-based model for recognizing multiple objects in images. The proposed model is a deep recurrent neural network trained with reinforcement learning to attend to the most relevant regions of the input image. We show that the model learns to both localize and recognize multiple objects despite bei...
[]
null
16
1412.7755
iclr_archive
[ 0.017669981345534325, -0.009857694618403912, -0.0006785523146390915, 0.038145486265420914, 0.02082555741071701, 0.052912306040525436, 0.016819069162011147, 0.04299909248948097, -0.054102737456560135, -0.0336168073117733, -0.04312604293227196, 0.023861344903707504, -0.07752344012260437, 0.0...
Deep Narrow Boltzmann Machines are Universal Approximators
https://arxiv.org/abs/1411.3784
[ "Guido Montufar" ]
Poster
null
We show that deep narrow Boltzmann machines are universal approximators of probability distributions on the activities of their visible units, provided they have sufficiently many hidden layers, each containing the same number of units as the visible layer. We show that, within certain parameter domains, deep Boltzma...
[]
null
17
1411.3784
iclr_archive
[ -0.006936711259186268, -0.01011824794113636, 0.024661386385560036, 0.011755472049117088, 0.027356121689081192, 0.028804650530219078, 0.03377220407128334, -0.006372632458806038, -0.027808982878923416, -0.009756515733897686, -0.005034138914197683, -0.0019038805039599538, -0.06233271583914757, ...
Transformation Properties of Learned Visual Representations
https://arxiv.org/abs/1412.7659
[ "Taco Cohen", "Max Welling" ]
Poster
null
When a three-dimensional object moves relative to an observer, a change occurs on the observer's image plane and in the visual representation computed by a learned model. Starting with the idea that a good visual representation is one that transforms linearly under scene motions, we show, using the theory of group re...
[]
null
18
1412.7659
iclr_archive
[ -0.013971688225865364, 0.015113377943634987, 0.023091034963726997, 0.02716953121125698, 0.026847993955016136, 0.012793569825589657, 0.027581004425883293, 0.021544551476836205, -0.047502439469099045, -0.03313093259930611, -0.03780336678028107, -0.011844652704894543, -0.0672769844532013, 0.0...
Joint RNN-Based Greedy Parsing and Word Composition
https://arxiv.org/abs/1412.7028
[ "Joël Legrand", "Ronan Collobert" ]
Poster
null
This paper introduces a greedy parser based on neural networks, which leverages a new compositional sub-tree representation. The greedy parser and the compositional procedure are jointly trained, and tightly depends on each-other. The composition procedure outputs a vector representation which summarizes syntacticall...
[]
null
19
1412.7028
iclr_archive
[ -0.024337265640497208, -0.008996652439236641, -0.01224612258374691, 0.030257005244493484, 0.025991110131144524, 0.05780455097556114, -0.0018762541003525257, 0.016837436705827713, -0.029729753732681274, -0.016626697033643723, -0.021017983555793762, 0.007862107828259468, -0.05321726202964783, ...
Adam: A Method for Stochastic Optimization
https://arxiv.org/abs/1412.6980
[ "Jimmy Ba", "Diederik Kingma" ]
Poster
null
We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradien...
[]
null
20
1412.6980
iclr_archive
[ -0.03645003214478493, -0.019249575212597847, 0.0092319892719388, 0.02856167033314705, 0.009572635404765606, 0.07320912927389145, 0.026167605072259903, 0.01353389024734497, 0.005823054350912571, -0.035289742052555084, -0.018081869930028915, 0.015893155708909035, -0.06424727290868759, -0.020...
Neural Machine Translation by Jointly Learning to Align and Translate
https://arxiv.org/abs/1409.0473
[ "Dzmitry Bahdanau", "Kyunghyun Cho", "Yoshua Bengio" ]
Poster
null
Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural ...
[]
null
21
1409.0473
iclr_archive
[ -0.005684528965502977, -0.03467094525694847, -0.007944873534142971, 0.018774930387735367, 0.016290921717882156, 0.06396574527025223, 0.015409497544169426, 0.02249527908861637, -0.01750517264008522, -0.04119409993290901, -0.031601957976818085, 0.03298115357756615, -0.03783518448472023, 0.00...
Scheduled denoising autoencoders
https://arxiv.org/abs/1406.3269
[ "Krzysztof Geras", "Charles Sutton" ]
Poster
null
We present a representation learning method that learns features at multiple different levels of scale. Working within the unsupervised framework of denoising autoencoders, we observe that when the input is heavily corrupted during training, the network tends to learn coarse-grained features, whereas when the input i...
[]
null
22
1406.3269
iclr_archive
[ -0.006281989626586437, -0.006547275930643082, -0.01581907644867897, 0.06447847187519073, 0.024005090817809105, 0.055926620960235596, 0.035407863557338715, -0.016236962750554085, -0.02409452013671398, -0.04124169796705246, 0.0017065302236005664, -0.007985716685652733, -0.059284407645463943, ...
Embedding Entities and Relations for Learning and Inference in Knowledge Bases
https://arxiv.org/abs/1412.6575
[ "Bishan Yang", "Scott Yih", "Xiaodong He", "Jianfeng Gao", "Li Deng" ]
Poster
null
We consider learning representations of entities and relations in KBs using the neural-embedding approach. We show that most existing models, including NTN (Socher et al., 2013) and TransE (Bordes et al., 2013b), can be generalized under a unified learning framework, where entities are low-dimensional vectors learned...
[]
null
23
1412.6575
iclr_archive
[ -0.01601749286055565, -0.007679739035665989, 0.020291930064558983, 0.03603529557585716, 0.049250248819589615, -0.008210496045649052, 0.01992804743349552, -0.01877010613679886, 0.007368351332843304, -0.006608247756958008, -0.022005127742886543, 0.03403407707810402, -0.06507959216833115, 0.0...
The local low-dimensionality of natural images
https://arxiv.org/abs/1412.6626
[ "Olivier Henaff", "Johannes Balle", "Neil Rabinowitz", "Eero Simoncelli" ]
Poster
null
We develop a new statistical model for photographic images, in which the local responses of a bank of linear filters are described as jointly Gaussian, with zero mean and a covariance that varies slowly over spatial position. We optimize sets of filters so as to minimize the nuclear norms of matrices of their local a...
[]
null
24
1412.6626
iclr_archive
[ -0.0010592426406219602, -0.0028493625577539206, 0.016646448522806168, 0.0419737882912159, 0.0419563390314579, 0.04619161784648895, 0.008415493182837963, -0.019615905359387398, -0.06176786497235298, -0.07221227884292603, -0.006884079892188311, -0.010244947858154774, -0.0753149539232254, -0....
Explaining and Harnessing Adversarial Examples
https://arxiv.org/abs/1412.6572
[ "Ian Goodfellow", "Jon Shlens", "Christian Szegedy" ]
Poster
null
Several machine learning models, including neural networks, consistently misclassify adversarial examples---inputs formed by applying small but intentionally worst-case perturbations to examples from the dataset, such that the perturbed input results in the model outputting an incorrect answer with high confidence. E...
[]
null
25
1412.6572
iclr_archive
[ -0.01907847635447979, -0.026390191167593002, -0.02593112923204899, 0.05809163674712181, 0.031651947647333145, 0.014792266301810741, 0.023291384801268578, -0.018169589340686798, -0.03261007368564606, -0.028481515124440193, -0.020135074853897095, 0.011972925625741482, -0.06567983329296112, -...
Modeling Compositionality with Multiplicative Recurrent Neural Networks
https://arxiv.org/abs/1412.6577
[ "Ozan Irsoy", "Claire Cardie" ]
Poster
null
We present the multiplicative recurrent neural network as a general model for compositional meaning in language, and evaluate it on the task of fine-grained sentiment analysis. We establish a connection to the previously investigated matrix-space models for compositionality, and show they are special cases of the mul...
[]
null
26
1412.6577
iclr_archive
[ 0.005548505578190088, -0.011056256480515003, -0.0036014732904732227, 0.036176037043333054, 0.034429922699928284, 0.0648011639714241, 0.01465521939098835, 0.04377050697803497, -0.01768777333199978, -0.010786251164972782, -0.02656627632677555, -0.013055704534053802, -0.0708191767334938, -0.0...
Very Deep Convolutional Networks for Large-Scale Image Recognition
https://arxiv.org/abs/1409.1556
[ "Karen Simonyan", "Andrew Zisserman" ]
Poster
null
In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improve...
[]
null
27
1409.1556
iclr_archive
[ 0.01817971095442772, -0.0473996140062809, 0.004819185007363558, 0.02936873771250248, 0.03162720799446106, 0.015696119517087936, 0.008704915642738342, 0.022175777703523636, -0.01362753938883543, -0.054987773299217224, 0.0107796099036932, -0.020215651020407677, -0.07133769243955612, 0.024413...
Speeding-up Convolutional Neural Networks Using Fine-tuned CP-Decomposition
https://arxiv.org/abs/1412.6553
[ "Vadim Lebedev", "Yaroslav Ganin", "Victor Lempitsky", "Maksim Rakhuba", "Ivan Oseledets" ]
Poster
null
We propose a simple two-step approach for speeding up convolution layers within large convolutional neural networks based on tensor decomposition and discriminative fine-tuning. Given a layer, we use non-linear least squares to compute a low-rank CP-decomposition of the 4D convolution kernel tensor into a sum of a sm...
[]
null
28
1412.6553
iclr_archive
[ -0.00957739818841219, -0.033649295568466187, 0.013097839429974556, 0.047072310000658035, 0.03745291382074356, 0.02238502912223339, -0.02201426401734352, -0.004342054482549429, -0.013786479830741882, -0.043794501572847366, -0.02982078678905964, -0.004548764321953058, -0.05921582132577896, 0...
Deep Captioning with Multimodal Recurrent Neural Networks (m-RNN)
https://arxiv.org/abs/1412.6632
[ "Junhua Mao", "Wei Xu", "Yi Yang", "Jiang Wang", "Alan Yuille" ]
Poster
null
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-netw...
[]
null
29
1412.6632
iclr_archive
[ -0.009589101187884808, -0.041977301239967346, -0.015281761065125465, 0.07077564299106598, 0.03211687132716179, 0.03752228990197182, 0.00904516689479351, 0.0399647057056427, -0.05049708113074303, -0.01798888109624386, -0.03508518263697624, 0.029002578929066658, -0.05790695920586586, -0.0007...
Deep Structured Output Learning for Unconstrained Text Recognition
https://arxiv.org/abs/1412.5903
[ "Max Jaderberg", "Karen Simonyan", "Andrea Vedaldi", "Andrew Zisserman" ]
Poster
null
We develop a representation suitable for the unconstrained recognition of words in natural images: the general case of no fixed lexicon and unknown length. To this end we propose a convolutional neural network (CNN) based architecture which incorporates a Conditional Random Field (CRF) graphical model, taking the w...
[]
null
30
1412.5903
iclr_archive
[ -0.00413596211001277, -0.027735279873013496, -0.004071063827723265, 0.0515449196100235, 0.033686406910419464, 0.029282517731189728, -0.0011436325730755925, 0.03766223043203354, 0.0021310923621058464, -0.02687574177980423, -0.027515815570950508, 0.033219192177057266, -0.0742066279053688, -0...
Zero-bias autoencoders and the benefits of co-adapting features
https://arxiv.org/abs/1402.3337
[ "Kishore Konda", "Roland Memisevic", "David Krueger" ]
Poster
null
Regularized training of an autoencoder typically results in hidden unit biases that take on large negative values. We show that negative biases are a natural result of using a hidden layer whose responsibility is to both represent the input data and act as a selection mechanism that ensures sparsity of the representa...
[]
null
31
1402.3337
iclr_archive
[ -0.003283413592725992, -0.009918085299432278, -0.0168695617467165, 0.0492064394056797, 0.030374540016055107, 0.05497258901596069, 0.020505541935563087, -0.01626710221171379, -0.0033847021404653788, -0.053557757288217545, -0.0069633303210139275, 0.022366123273968697, -0.08072246611118317, 0...
Automatic Discovery and Optimization of Parts for Image Classification
https://arxiv.org/abs/1412.6598
[ "Sobhan Naderi Parizi", "Andrea Vedaldi", "Andrew Zisserman", "Pedro Felzenszwalb" ]
Poster
null
Part-based representations have been shown to be very useful for image classification. Learning part-based models is often viewed as a two-stage problem. First, a collection of informative parts is discovered, using heuristics that promote part distinctiveness and diversity, and then classifiers are trained on the ve...
[]
null
32
1412.6598
iclr_archive
[ 0.015468486584722996, -0.01600276678800583, 0.0019417139701545238, 0.049846913665533066, 0.018617166206240654, 0.04272107779979706, -0.014966122806072235, -0.01917664147913456, -0.03426944464445114, -0.033994968980550766, -0.056602105498313904, -0.008623726665973663, -0.055353425443172455, ...
Understanding Locally Competitive Networks
https://arxiv.org/abs/1410.1165
[ "Rupesh Srivastava", "Jonathan Masci", "Faustino Gomez", "Juergen Schmidhuber" ]
Poster
null
Recently proposed neural network activation functions such as rectified linear, maxout, and local winner-take-all have allowed for faster and more effective training of deep neural architectures on large and complex datasets. The common trait among these functions is that they implement local competition between smal...
[]
null
33
1410.1165
iclr_archive
[ -0.04629231616854668, -0.02736620046198368, -0.014420234598219395, 0.03352563455700874, 0.04616902768611908, 0.03078104555606842, 0.004144294187426567, -0.012190531007945538, -0.020622849464416504, -0.020129309967160225, 0.026681847870349884, 0.00629772013053298, -0.050186313688755035, -0....
Leveraging Monolingual Data for Crosslingual Compositional Word Representations
https://arxiv.org/abs/1412.6334
[ "Hubert Soyer", "Pontus Stenetorp", "Akiko Aizawa" ]
Poster
null
In this work, we present a novel neural network based architecture for inducing compositional crosslingual word representations. Unlike previously proposed methods, our method fulfills the following three criteria; it constrains the word-level representations to be compositional, it is capable of leveraging both bili...
[]
null
34
1412.6334
iclr_archive
[ -0.004371400456875563, -0.011180587112903595, 0.006720326840877533, 0.03989139944314957, 0.021813038736581802, 0.026827797293663025, 0.00294155883602798, 0.027795230969786644, -0.009414936415851116, -0.021311862394213676, -0.019789498299360275, 0.010546883568167686, -0.0493544302880764, -0...
Move Evaluation in Go Using Deep Convolutional Neural Networks
https://arxiv.org/abs/1412.6564
[ "Chris Maddison", "Aja Huang", "Ilya Sutskever", "David Silver" ]
Poster
null
The game of Go is more challenging than other board games, due to the difficulty of constructing a position or move evaluation function. In this paper we investigate whether deep convolutional networks can be used to directly represent and learn this knowledge. We train a large 12-layer convolutional neural network b...
[]
null
35
1412.6564
iclr_archive
[ -0.03206796199083328, -0.049488164484500885, 0.00029614329105243087, 0.042780227959156036, 0.045022062957286835, 0.0008391279843635857, 0.009884670376777649, 0.03565339744091034, 0.004090771544724703, -0.036135800182819366, 0.01486896350979805, -0.005711401347070932, -0.06853080540895462, ...
Fast Convolutional Nets With fbfft: A GPU Performance Evaluation
https://arxiv.org/abs/1412.7580
[ "Nicolas Vasilache", "Jeff Johnson", "Michael Mathieu", "Soumith Chintala", "Serkan Piantino", "Yann LeCun" ]
Poster
null
We examine the performance profile of Convolutional Neural Network training on the current generation of NVIDIA Graphics Processing Units. We introduce two new Fast Fourier Transform convolution implementations: one based on NVIDIA's cuFFT library, and another based on a Facebook authored FFT implementation, fbfft, t...
[]
null
36
1412.7580
iclr_archive
[ 0.005991718731820583, -0.04726449027657509, 0.025601275265216827, 0.025315532460808754, 0.03517220914363861, 0.03579822927713394, -0.012910126708447933, 0.049497686326503754, -0.007204278372228146, -0.05632399022579193, 0.015573403798043728, 0.0014487336156889796, -0.07538507133722305, -0....
Word Representations via Gaussian Embedding
https://arxiv.org/abs/1412.6623
[ "Luke Vilnis", "Andrew McCallum" ]
Poster
null
Current work in lexical distributed representations maps each word to a point vector in low-dimensional space. Mapping instead to a density provides many interesting advantages, including better capturing uncertainty about a representation and its relationships, expressing asymmetries more naturally than dot product ...
[]
null
37
1412.6623
iclr_archive
[ -0.022525973618030548, -0.007958680391311646, 0.0029189216438680887, 0.06130314990878105, 0.026897192001342773, 0.05592765286564827, 0.02717588283121586, 0.0026656747795641422, -0.002198993694037199, -0.03068809024989605, -0.009585809893906116, 0.015337381511926651, -0.07212560623884201, 0...
Qualitatively characterizing neural network optimization problems
https://arxiv.org/abs/1412.6544
[ "Ian Goodfellow", "Oriol Vinyals" ]
Poster
null
Training neural networks involves solving large-scale non-convex optimization problems. This task has long been believed to be extremely difficult, with fear of local minima and other obstacles motivating a variety of schemes to improve optimization, such as unsupervised pretraining. However, modern neural networks a...
[]
null
38
1412.6544
iclr_archive
[ -0.04026986286044121, -0.022642264142632484, -0.005983470473438501, 0.04811333492398262, 0.03697055205702782, 0.050341468304395676, -0.0063139148987829685, -0.016202392056584358, -0.03586403653025627, -0.03219089284539223, -0.019967157393693924, 0.022246871143579483, -0.0397377647459507, 0...
Memory Networks
https://arxiv.org/abs/1410.3916
[ "Jason Weston", "Sumit Chopra", "Antoine Bordes" ]
Poster
null
We describe a new class of learning models called memory networks. Memory networks reason with inference components combined with a long-term memory component; they learn how to use these jointly. The long-term memory can be read and written to, with the goal of using it for prediction. We investigate these models in...
[]
null
39
1410.3916
iclr_archive
[ -0.009670330211520195, 0.001341706607490778, -0.009222445078194141, 0.0511188805103302, 0.04864468052983284, 0.01231493428349495, -0.005306434817612171, 0.021064434200525284, -0.0411897748708725, 0.010568782687187195, 0.0015729879960417747, 0.024641918018460274, -0.04645440727472305, -0.01...
Generative Modeling of Convolutional Neural Networks
https://arxiv.org/abs/1412.6296
[ "Jifeng Dai", "Yang Lu", "Ying-Nian Wu" ]
Poster
null
The convolutional neural networks (CNNs) have proven to be a powerful tool for discriminative learning. Recently researchers have also started to show interest in the generative aspects of CNNs in order to gain a deeper understanding of what they have learned and how to further improve them. This paper investigates g...
[]
null
40
1412.6296
iclr_archive
[ 0.021600598469376564, -0.023329369723796844, -0.0077382950112223625, 0.06096671521663666, 0.024356914684176445, 0.038858845829963684, -0.01665736362338066, 0.013275419361889362, -0.026230674237012863, -0.05039505288004875, -0.013955784030258656, -0.02049321122467518, -0.059901464730501175, ...
A Unified Perspective on Multi-Domain and Multi-Task Learning
https://arxiv.org/abs/1412.7489
[ "Yongxin Yang", "Timothy Hospedales" ]
Poster
null
In this paper, we provide a new neural-network based perspective on multi-task learning (MTL) and multi-domain learning (MDL). By introducing the concept of a semantic descriptor, this framework unifies MDL and MTL as well as encompassing various classic and recent MTL/MDL algorithms by interpreting them as different...
[]
null
41
1412.7489
iclr_archive
[ 0.008415445685386658, 0.0007903411169536412, -0.015737637877464294, 0.04130909591913223, 0.03519966080784798, 0.03630171716213226, 0.0217235516756773, 0.004557198844850063, -0.004807495977729559, -0.020902160555124283, -0.023371919989585876, 0.037543702870607376, -0.0780433863401413, -0.00...
Object detectors emerge in Deep Scene CNNs
https://arxiv.org/abs/1412.6856
[ "Bolei Zhou", "Aditya Khosla", "Agata Lapedriza", "Aude Oliva", "Antonio Torralba" ]
Poster
null
With the success of new computational architectures for visual processing, such as convolutional neural networks (CNN) and access to image databases with millions of labeled examples (e.g., ImageNet, Places), the state of the art in computer vision is advancing rapidly. One important factor for continued progress is ...
[]
null
42
1412.6856
iclr_archive
[ -0.006357009056955576, 0.001259227399714291, 0.02100524492561817, 0.04599931463599205, 0.024247700348496437, 0.010110636241734028, 0.010036276653409004, 0.01003933697938919, -0.04300607740879059, -0.03792363777756691, -0.03891061991453171, 0.006014237646013498, -0.05411538854241371, -0.002...