id stringlengths 11 20 | paper_text stringlengths 29 163k | review stringlengths 666 24.3k |
|---|---|---|
iclr_2018_B1EVwkqTW | While deep neural networks have shown outstanding results in a wide range of applications, learning from a very limited number of examples is still a challenging task. Despite the difficulties of the few-shot learning, metric-learning techniques showed the potential of the neural networks for this task. While these met... | Make SVM great again with Siamese kernel for few-shot learning
** PAPER SUMMARY **
The author proposes to combine siamase networks with an SVM for pair classification. The proposed approach is evaluated on few shot learning tasks, on omniglot and timit.
** REVIEW SUMMARY **
The paper is readable but it could be more... |
iclr_2018_SJTB5GZCb | The biological plausibility of the backpropagation algorithm has long been doubted by neuroscientists. Two major reasons are that neurons would need to send two different types of signal in the forward and backward phases, and that pairs of neurons would need to communicate through symmetric bidirectional connections. ... | The manuscript discusses a learning algorithm that is based on the equilibrium propagation method, which can be applied to networks with asymmetric connections. This extension is interesting, but the results seem to be incomplete and missing necessary additional analyses. Therefore, I do not recommend acceptance of the... |
iclr_2018_SJIA6ZWC- | Machine learning models are usually tuned by nesting optimization of model weights inside the optimization of hyperparameters. We give a method to collapse this nested optimization into joint stochastic optimization of both weights and hyperparameters. Our method trains a neural network to output approximately optimal ... | *Summary*
The paper proposes to use hyper-networks [Ha et al. 2016] for the tuning of hyper-parameters, along the lines of [Brock et al. 2017]. The core idea is to have a side neural network sufficiently expressive to learn the (large-scale, matrix-valued) mapping from a given configuration of hyper-parameters to the w... |
iclr_2018_Hyg0vbWC- | Published as a conference paper at ICLR 2018 GENERATING WIKIPEDIA BY SUMMARIZING LONG SEQUENCES
We show that generating English Wikipedia articles can be approached as a multidocument summarization of source documents. We use extractive summarization to coarsely identify salient information and a neural abstractive mod... | The main significance of this paper is to propose the task of generating the lead section of Wikipedia articles by viewing it as a multi-document summarization problem. Linked articles as well as the results of an external web search query are used as input documents, from which the Wikipedia lead section must be gener... |
iclr_2018_ByQZjx-0- | We propose Efficient Neural Architecture Search (ENAS), a faster and less expensive approach to automated model design than previous methods. In ENAS, a controller learns to discover neural network architectures by searching for an optimal path within a larger model. The controller is trained with policy gradient to se... | Summary:
The paper presents a method for learning certain aspects of a neural network architecture, specifically the number of output maps in certain connections and the existence of skip connections. The method is relatively efficient, since it searches in a space of similar architectures, and uses weights sharing be... |
iclr_2018_HyyP33gAZ | Published as a conference paper at ICLR 2018 ACTIVATION MAXIMIZATION GENERATIVE ADVER- SARIAL NETS
Class labels have been empirically shown useful in improving the sample quality of generative adversarial nets (GANs). In this paper, we mathematically study the properties of the current variants of GANs that make use of... | This paper is a thorough investigation of various “class aware” GAN architectures. It purposes a variety of modifications on existing approaches and additionally provides extensive analysis of the commonly used Inception Score evaluation metric.
The paper starts by introducing and analyzing two previous class aware GAN... |
iclr_2018_HyI6s40a- | Recent advances in adversarial Deep Learning (DL) have opened up a new and largely unexplored surface for malicious attacks jeopardizing the integrity of autonomous DL systems. We introduce a novel automated countermeasure called Parallel Checkpointing Learners (PCL) to thwart the potential adversarial attacks and sign... | Summary:
The paper presents an unsupervised method for detecting adversarial examples of neural networks. The method includes two independent components: an ‘input defender’ which tried to inspect the input, and a ‘latent defender’ trying to inspect a hidden representation. Both are based on the claim that adversarial... |
iclr_2018_Sy-dQG-Rb | Published as a conference paper at ICLR 2018 NEURAL SPEED READING VIA SKIM-RNN
Inspired by the principles of speed reading, we introduce Skim-RNN, a recurrent neural network (RNN) that dynamically decides to update only a small fraction of the hidden state for relatively unimportant input tokens. Skim-RNN gives computa... | Summary: The paper proposes a learnable skimming mechanism for RNN. The model decides whether to send the word to a larger heavy-weight RNN or a light-weight RNN. The heavy-weight and the light-weight RNN each controls a portion of the hidden state. The paper finds that with the proposed skimming method, they achieve a... |
iclr_2018_HkPCrEZ0Z | Model-free deep reinforcement learning algorithms are able to successfully solve a wide range of continuous control tasks, but typically require many on-policy samples to achieve good performance. Model-based RL algorithms are sampleefficient on the other hand, while learning accurate global models of complex dynamic e... | This paper presents a model-based approach to variance reduction in policy gradient methods. The basic idea is to use a multi-step dynamics model as a "baseline" (more properly a control variate, as the terminology in the paper uses, but I think baselines are more familiar to the RL community) to reduce the variance o... |
iclr_2018_ryZElGZ0Z | The ability of an agent to discover its own learning objectives has long been considered a key ingredient for artificial general intelligence. Breakthroughs in autonomous decision making and reinforcement learning have primarily been in domains where the agent's goal is outlined and clear: such as playing a game to win... | I really enjoyed reading this paper and stopped a few time to write down new ideas it brought up. Well written and very clear, but somewhat lacking in the experimental or theoretical results.
The formulation of AdaGain is very reminiscent of the SGA algorithm in Kushner & Yin (2003), and more generally gradient descent... |
iclr_2018_BkrSv0lA- | LOSS-AWARE WEIGHT QUANTIZATION OF DEEP NET- WORKS
The huge size of deep networks hinders their use in small computing devices. In this paper, we consider compressing the network by weight quantization. We extend a recently proposed loss-aware weight binarization scheme to ternarization, with possibly different scaling ... | This paper proposes a new method to train DNNs with quantized weights, by including the quantization as a constraint in a proximal quasi-Newton algorithm, which simultaneously learns a scaling for the quantized values (possibly different for positive and negative weights).
The paper is very clearly written, and the pr... |
iclr_2018_B1Gi6LeRZ | Published as a conference paper at ICLR 2018 LEARNING FROM BETWEEN-CLASS EXAMPLES FOR DEEP SOUND RECOGNITION
Deep learning methods have achieved high performance in sound recognition tasks. Deciding how to feed the training data is important for further performance improvement. We propose a novel learning method for de... | This manuscript proposes a method to improve the performance of a generic learning method by generating "in between class" (BC) training samples. The manuscript motivates the necessity of such technique and presents the basic intuition. The authors show how the so-called BC learning helps training different deep archit... |
iclr_2018_BJypUGZ0Z | Workshop track -ICLR 2018 ACCELERATING NEURAL ARCHITECTURE SEARCH US- ING PERFORMANCE PREDICTION
Methods for neural network hyperparameter optimization and architecture search are computationally expensive due to the need to train a large number of model configurations. In this paper, we show that simple regression mod... | This paper shows a simple method for predicting the performance that neural networks will achieve with a given architecture, hyperparameters, and based on an initial part of the learning curve.
The method assumes that it is possible to first execute 100 evaluations up to the total number of epochs.
From these 100 evalu... |
iclr_2018_BJgPCveAW | We propose a novel way of reducing the number of parameters in the storagehungry fully connected layers of a neural network by using pre-defined sparsity, where the majority of connections are absent prior to starting training. Our results indicate that convolutional neural networks can operate without any loss of accu... | This paper examines sparse connection patterns in upper layers of convolutional image classification networks. Networks with very few connections in the upper layers are experimentally determined to perform almost as well as those with full connection masks. Heuristics for distributing connections among windows/group... |
iclr_2018_Hk6WhagRW | Published as a conference paper at ICLR 2018 EMERGENT COMMUNICATION THROUGH NEGOTIATION
Multi-agent reinforcement learning offers a way to study how communication could emerge in communities of agents needing to solve specific problems. In this paper, we study the emergence of communication in the negotiation environme... | The authors describe a variant of the negotiation game in which agents of different type, selfish or prosocial, and with different preferences. The central feature is the consideration of a secondary communication (linguistic) channel for the purpose of cheap talk, i.e. talk whose semantics are not laid out a priori.
... |
iclr_2018_B1X0mzZCW | Published as a conference paper at ICLR 2018 FIDELITY-WEIGHTED LEARNING
Training deep neural networks requires many training samples, but in practice training labels are expensive to obtain and may be of varying quality, as some may be from trusted expert labelers while others might be from heuristics or other sources ... | The problem of interest is to train deep neural network models with few labelled training samples. The specific assumption is there is a large pool of unlabelled data, and a heuristic function that can provide label annotations, possibly with varying levels of noises, to those unlabelled data. The adopted learning mode... |
iclr_2018_S1Y7OOlRZ | Modern machine learning models are characterized by large hyperparameter search spaces and prohibitively expensive training costs. For such models, we cannot afford to train candidate models sequentially and wait months before finding a suitable hyperparameter configuration. Hence, we introduce the large-scale regime f... | This paper introduces a simple extension to parallelize Hyperband.
Points in favor of the paper:
* Addresses an important problem
Points against:
* Only 5-fold speedup by parallelization with 5 x 25 workers, and worse performance in the same budget than Google Vizier (even though that treats the problem as a black box... |
iclr_2018_rJwelMbR- | DIVIDE-AND-CONQUER REINFORCEMENT LEARNING
Standard model-free deep reinforcement learning (RL) algorithms sample a new initial state for each trial, allowing them to optimize policies that can perform well even in highly stochastic environments. However, problems that exhibit considerable initial state variation typica... | This paper presents a method for learning a global policy over multiple different MDPs (referred to as different "contexts", each MDP having the same dynamics and reward, but different initial state). The basic idea is to learn a separate policy for each context, but regularized in a manner that keeps all of them rela... |
iclr_2018_rkcya1ZAW | Two fundamental problems in unsupervised learning are efficient inference for latent-variable models and robust density estimation based on large amounts of unlabeled data. For efficient inference, normalizing flows have been recently developed to approximate a target distribution arbitrarily well. In practice, however... | The authors propose the use of first order Langevin dynamics as a way to transition from one latent variable to the next in the VAE setting, as opposed to the deterministic transitions of normalizing flow. The extremely popular Fokker-Planck equation is used to analyze the steady state distributions in this setting. Th... |
iclr_2018_S1CChZ-CZ | Published as a conference paper at ICLR 2018 ASK THE RIGHT QUESTIONS: ACTIVE QUESTION REFORMULATION WITH REINFORCEMENT LEARNING
We frame Question Answering (QA) as a Reinforcement Learning task, an approach that we call Active Question Answering. We propose an agent that sits between the user and a black box QA system ... | This paper formulates the Jeopardy QA as a query reformulation task that leverages a search engine. In particular, a user will try a sequence of alternative queries based on the original question in order to find the answer. The RL formulation essentially tries to mimic this process. Although this is an interesting ... |
iclr_2018_Hk6kPgZA- | Published as a conference paper at ICLR 2018 CERTIFYING SOME DISTRIBUTIONAL ROBUSTNESS WITH PRINCIPLED ADVERSARIAL TRAINING
Neural networks are vulnerable to adversarial examples and researchers have proposed many heuristic attack and defense mechanisms. We address this problem through the principled lens of distributi... | This paper proposes a principled methodology to induce distributional robustness in trained neural nets with the purpose of mitigating the impact of adversarial examples. The idea is to train the model to perform well not only with respect to the unknown population distribution, but to perform well on the worst-case di... |
iclr_2018_HJ39YKiTb | In this paper, we propose the Associative Conversation Model that generates visual information from textual information and uses it for generating sentences in order to utilize visual information in a dialogue system without image input. In research on Neural Machine Translation, there are studies that generate transla... | The paper proposes to augment (traditional) text-based sentence generation/dialogue approaches by incorporating visual information. The idea is that associating visual information with input text, and using that associated visual information as additional input will produce better output text than using only the origi... |
iclr_2018_HJPSN3gRW | In this work, we focus on the problem of grounding language by training an agent to follow a set of natural language instructions and navigate to a target object in a 2D grid environment. The agent receives visual information through raw pixels and a natural language instruction telling what task needs to be achieved. ... | **Paper Summary**
The paper studies the problem of navigating to a target object in a 2D grid environment by following given natural language description as well as receiving visual information as raw pixels. The proposed architecture consists of a convoutional neural network encoding visual input, gated recurrent uni... |
iclr_2018_Sy-tszZRZ | In this paper, we study the representational power of deep neural networks (DNN) that belong to the family of piecewise-linear (PWL) functions, based on PWL activation units such as rectifier or maxout. We investigate the complexity of such networks by studying the number of linear regions of the PWL function. Typicall... | Paper Summary:
This paper looks at providing better bounds for the number of linear regions in the function represented by a deep neural network. It first recaps some of the setting: if a neural network has a piecewise linear activation function (e.g. relu, maxout), the final function computed by the network (before so... |
iclr_2018_B1lMMx1CW | Workshop track -ICLR 2018 THE EFFECTIVENESS OF A TWO-LAYER NEURAL NET- WORK FOR RECOMMENDATIONS
We present a personalized recommender system using neural network for recommending products, such as eBooks, audio-books, Mobile Apps, Video and Music. It produces recommendations based on customer's implicit feedback histor... | The paper proposes a new neural network based method for recommendation.
The main finding of the paper is that a relatively simple method works for recommendation, compared to other methods based on neural networks that have been recently proposed.
This contribution is not bad for an empirical paper. There's certainly ... |
iclr_2018_r1BRfhiab | We consider neural network training, in applications in which there are many possible classes, but at test-time, the task is to identify only whether the given example belongs to a specific class, which can be different in different applications of the classifier. For instance, this is the case in an image search engin... | The paper addresses the problem of a mismatch between training classification loss and a loss at test time. This is motivated by use cases in which multiclass classification problems are learned during training, but where binary or reduced multi-class classifications is performed at test time. The question for me is th... |
iclr_2018_SJJySbbAZ | TRAINING GANS WITH OPTIMISM
We address the issue of limit cycling behavior in training Generative Adversarial Networks and propose the use of Optimistic Mirror Decent (OMD) for training Wasserstein GANs. Recent theoretical results have shown that optimistic mirror decent (OMD) can enjoy faster regret rates in the conte... | This paper proposes the use of optimistic mirror descent to train Wasserstein Generative Adversarial Networks (WGANS). The authors remark that the current training of GANs, which amounts to solving a zero-sum game between a generator and discriminator, is often unstable, and they argue that one source of instability is... |
iclr_2018_S1Auv-WRZ | Effective training of neural networks requires much data. In the low-data regime, parameters are underdetermined, and learnt networks generalise poorly. Data Augmentation (Krizhevsky et al., 2012) alleviates this by using existing data more effectively. However standard data augmentation produces only limited plausible... | This paper proposes a conditional Generative Adversarial Networks that is used for data augmentation. In order to evaluate the performance of the proposed model, they use Omniglot, EMNIST, and VGG-Faces datasets and uses in the meta-learning task and standard classification task in the low-data regime. The paper is wel... |
iclr_2018_HyI5ro0pW | Artificial neural networks have opened up a world of possibilities in data science and artificial intelligence, but neural networks are cumbersome tools that grow with the complexity of the learning problem. We make contributions to this issue by considering a modified version of the fully connected layer we call a blo... | The paper proposes to make the inner layers in a neural network be block diagonal, mainly as an alternative to pruning. The implementation of this seems straightforward, and can be done either via initialization or via pruning on the off-diagonals. There are a few ideas the paper discusses:
(1) compared to pruning weig... |
iclr_2018_rylejExC- | Graph convolutional networks (GCNs) are powerful deep neural networks for graph-structured data. However, GCN computes nodes' representation recursively from their neighbors, making the receptive field size grow exponentially with the number of layers. Previous attempts on reducing the receptive field size by subsampli... | The paper proposes a method to speed up the training of graph convolutional networks, which are quite slow for large graphs. The key insight is to improve the estimates of the average neighbor activations (via neighbor sampling) so that we can either sample less neighbors or have higher accuracy for the same number of ... |
iclr_2018_HkuGJ3kCb | ALL-BUT-THE-TOP: SIMPLE AND EFFECTIVE POST- PROCESSING FOR WORD REPRESENTATIONS
Real-valued word representations have transformed NLP applications; popular examples are word2vec and GloVe, recognized for their ability to capture linguistic regularities. In this paper, we demonstrate a very simple, and yet counter-intui... | This paper proposes a simple post-processing technique for word representations designed to improve representational quality and performance on downstream tasks. The procedure involves mean subtraction followed by projecting out the first D principle directions and is motivated by improving isotropy of the partition fu... |
iclr_2018_S1pWFzbAW | The large memory requirements of deep neural networks limit their deployment and adoption on many devices. Model compression methods effectively reduce the memory requirements of these models, usually through applying transformations such as weight pruning or quantization. In this paper, we present a novel scheme for l... | This paper proposes an interesting approach to compress the weights of a network for storage or transmission purposes. My understanding is, at inference, the network is 'recovered' therefore there is no difference in processing time (slight differences in accuracy due to the approximation in recovering the weights).
- ... |
iclr_2018_SkAK2jg0b | Transfer learning for feature extraction can be used to exploit deep representations in contexts where there is very few training data, where there are limited computational resources, or when tuning the hyper-parameters needed for training is not an option. While previous contributions to feature extraction propose em... | The paper addresses the scenario when using a pretrained deep network as learnt feature representation for another (small) task where retraining is not an option or not desired. In this situation it proposes to use all layers of the network to extract feature from, instead of only one layer.
Then it proposes to standa... |
iclr_2018_SyKoKWbC- | In most current formulations of adversarial training, the discriminators can be expressed as single-input operators, that is, the mapping they define is separable over observations. In this work, we argue that this property might help explain the infamous mode collapse phenomenon in adversarially-trained generative mod... | The paper proposes to replace single-sample discriminators in adversarial training with discriminators that explicitly operate on distributions of examples, so as to incentivize the generator to cover the full distribution of the training data and not collapse to isolated modes.
The idea of avoiding mode collapse by p... |
iclr_2018_Hk0wHx-RW | Published as a conference paper at ICLR 2018 LEARNING SPARSE LATENT REPRESENTATIONS WITH THE DEEP COPULA INFORMATION BOTTLENECK
Deep latent variable models are powerful tools for representation learning. In this paper, we adopt the deep information bottleneck model, identify its shortcomings and propose a model that ci... | This paper identifies and proposes a fix for a shortcoming of the Deep Information Bottleneck approach, namely that the induced representation is not invariant to monotonic transform of the marginal distributions (as opposed to the mutual information on which it is based). The authors address this shortcoming by applyi... |
iclr_2018_BkDB51WR- | We propose to tackle a time series regression problem by computing temporal evolution of a probability density function to provide a probabilistic forecast. A Recurrent Neural Network (RNN) based model is employed to learn a nonlinear operator for temporal evolution of a probability density function. We use a softmax l... | Interesting ideas that extend LSTM to produce probabilistic forecasts for univariate time series, experiments are okay. Unclear if this would work at all in higher-dimensional time series. It is also unclear to me what are the sources of the uncertainties captured.
The author proposed to incorporate 2 different discret... |
iclr_2018_ry8dvM-R- | Published as a conference paper at ICLR 2018 ROUTING NETWORKS: ADAPTIVE SELECTION OF NON-LINEAR FUNCTIONS FOR MULTI-TASK LEARN- ING
Multi-task learning (MTL) with neural networks leverages commonalities in tasks to improve performance, but often suffers from task interference which reduces the benefits of transfer. To ... | Summary:
The paper suggests to use a modular network with a controller which makes decisions, at each time step, regarding the next nodule to apply. This network is suggested a tool for solving multi-task scenarios, where certain modules may be shared and others may be trained independently for each task. It is propose... |
iclr_2018_rJoXrxZAZ | This paper introduces HybridNet, a hybrid neural network to speed-up autoregressive models for raw audio waveform generation. As an example, we propose a hybrid model that combines an autoregressive network named WaveNet and a conventional LSTM model to address speech synthesis. Instead of generating one sample per tim... | This paper presents HybridNet, a neural speech (and other audio) synthesis system (vocoder) that combines the popular and effective WaveNet model with an LSTM with the goal of offering a model with faster inference-time audio generation.
Summary: The proposed model, HybridNet is a fairly straightforward variation of Wa... |
iclr_2018_rJSr0GZR- | Most deep latent factor models choose simple priors for simplicity, tractability or not knowing what prior to use. Recent studies show that the choice of the prior may have a profound effect on the expressiveness of the model, especially when its generative network has limited capacity. In this paper, we propose to lea... | This paper propose a simple extension of the adversarial auto-encoders for (conditional) image generation. The general idea is that instead of using Gaussian prior, the propose algorithm uses a "code generator" network to warp the gaussian distribution, such that the internal prior of the latent encoding space is more... |
iclr_2018_Syx6bz-Ab | Relational databases store a significant amount of the worlds data. However, accessing this data currently requires users to understand a query language such as SQL. We propose Seq2SQL, a deep neural network for translating natural language questions to corresponding SQL queries. Our model uses rewards from inthe-loop ... | This paper presents a new approach to support the conversion from natural language to database queries.
One of the major contributions of the work is the introduction of a new real-world benchmark dataset based on questions over Wikipedia. The scale of the data set is significantly larger than any existing ones. Howev... |
iclr_2018_Bk8ZcAxR- | EIGENOPTION DISCOVERY THROUGH THE DEEP SUCCESSOR REPRESENTATION
Options in reinforcement learning allow agents to hierarchically decompose a task into subtasks, having the potential to speed up learning and planning. However, autonomously learning effective sets of options is still a major challenge in the field. In th... | Eigenoption Discovery Through the Deep Successor Representation
The paper is a follow up on previous work by Machado et al. (2017) showing how proto-value functions (PVFs) can be used to define options called “eigenoptions”. In essence, Machado et al. (2017) showed that, in the tabular case, if you interpret the differ... |
iclr_2018_ryj0790hb | Given an existing trained neural network, it is often desirable to learn new capabilities without hindering performance of those already learned. Existing approaches either learn sub-optimal solutions, require joint training, or incur a substantial increment in the number of parameters for each added task, typically as... | This paper proposes to adapt convnet representations to new tasks while avoiding catastrophic forgetting by learning a per-task “controller” specifying weightings of the convolution-al filters throughout the network while keeping the filters themselves fixed.
Pros
The proposed approach is novel and broadly applicable. ... |
iclr_2018_SkfNU2e0Z | Building deep neural networks to control autonomous agents which have to interact in real-time with the physical world, such as robots or automotive vehicles, requires a seamless integration of time into a network's architecture. The central question of this work is, how the temporal nature of reality should be reflect... | This paper introduces a new toolbox for deep neural networks learning and evaluation. The central idea is to include time in the processing of all the units in the network. For this, the authors propose a paradigm switch: form layerwise-sequential networks, where at every time frame the network is evaluated by updating... |
iclr_2018_Sy0GnUxCb | Published as a conference paper at ICLR 2018 EMERGENT COMPLEXITY VIA MULTI-AGENT COMPETITION
Reinforcement learning algorithms can train agents that solve problems in complex, interesting environments. Normally, the complexity of the trained agent is closely related to the complexity of the environment. This suggests t... | In this paper, the authors produced quite cool videos showing the acquisition of highly complex skills, and they are happy about it. If you read the conclusion, this is the only message they put forward, and to me this is not a scientific message.
A more classical summary is that the authors use PPO, a state-of-the-art... |
iclr_2018_HJXOfZ-AZ | According to parallel distributed processing (PDP) theory in psychology, neural networks (NN) learn distributed rather than interpretable localist representations. This view has been held so strongly that few researchers have analysed single units to determine if this assumption is correct. However, recent results from... | This paper studies the development of localist representations in the hidden layers
of feed-forward neural networks.
The idea is interesting and the findings are intriguing. Local codes
increase understandability and could be important for better
understanding natural neural networks. Understanding how local codes
for... |
iclr_2018_B1CNpYg0- | Words in natural language follow a Zipfian distribution whereby some words are frequent but most are rare. Learning representations for words in the "long tail" of this distribution requires enormous amounts of data. Representations of rare words trained directly on end tasks are usually poor, requiring us to pre-train... | This paper examines ways of producing word embeddings for rare words on demand. The key real-world use case is for domain specific terms, but here the techniques are demonstrated on rarer words in standard data sets. The strength of this paper is that it both gives a more systematic framework for and builds on existing... |
iclr_2018_HJC2SzZCW | Published as a conference paper at ICLR 2018 SENSITIVITY AND GENERALIZATION IN NEURAL NETWORKS: AN EMPIRICAL STUDY
In practice it is often found that large over-parameterized neural networks generalize better than their smaller counterparts, an observation that appears to conflict with classical notions of function com... | This paper proposes an analysis of the robustness of deep neural networks with respect to data perturbations.
*Quality*
The quality of exposition is not satisfactory. Actually, the paper is pretty difficult to evaluate at the present stage and it needs a drastic change in the writing style.
*Clarity*
The paper is not ... |
iclr_2018_BkeC_J-R- | Reinforcement learning methods have recently achieved impressive results on a wide range of control problems. However, especially with complex inputs, they still require an extensive amount of training data in order to converge to a meaningful solution. This limitation largely prohibits their usage for complex input sp... | This paper proposes leveraging labelled controlled data to accelerate reinforcement-based learning of a control policy. It provides two main contributions: pre-training the policy network of a DDPG agent in a supervised manner so that it begins in reasonable state-action distribution and regalurizing the Q-updates of ... |
iclr_2018_SyMvJrdaW | DECOUPLING THE LAYERS IN RESIDUAL NETWORKS
We propose a Warped Residual Network (WarpNet) using a parallelizable warp operator for forward and backward propagation to distant layers that trains faster than the original residual neural network. We apply a perturbation theory on residual networks and decouple the interac... | The main contribution of this paper is a particular Taylor expansion of the outputs of a ResNet which is shown to be exact at almost all points in the input space. This expression is used to develop a new layer called a “warp layer” which essentially tries to compute several layers of the residual network using the Ta... |
iclr_2018_ByCPHrgCW | When deep learning is applied to sensitive data sets, many privacy-related implementation issues arise. These issues are especially evident in the healthcare, finance, law and government industries. Homomorphic encryption could allow a server to make inferences on inputs encrypted by a client, but to our best knowledge... | The paper presents a means of evaluating a neural network securely using homomorphic encryption. A neural network is already trained, and its weights are public. The network is to be evaluated over a private input, so that only the final outcome of the computation-and nothing but that-is finally learned.
The authors ta... |
iclr_2018_S1EzRgb0W | Neural networks make mistakes. The reason why a mistake is made often remains a mystery. It would be useful to have a method that can give an explanation that is intuitive to a user as to why an image is misclassified. In this paper we develop a method for explaining the mistakes of an image classification model by vis... | In this paper, the authors aim to better understand the classification of neural networks. The authors explore the latent space of a variational auto encoder and consider the perturbations of the latent space in order to obtain the correct classification. They evaluate their method on CelebA and MNIST datasets.
Pros:
... |
iclr_2018_ByJDAIe0b | Episodic memory is a psychology term which refers to the ability to recall specific events from the past. We suggest one advantage of this particular type of memory is the ability to easily assign credit to a specific state when remembered information is found to be useful. Inspired by this idea, and the increasing pop... | The paper proposes a modified approach to RL, where an additional "episodic memory" is kept by the agent. What this means is that the agent has a reservoir of n "states" in which states encountered in the past can be stored. There are then of course two main questions to address (i) which states should be stored and ho... |
iclr_2018_SyrGJYlRZ | Hyperparameter tuning is one of the most time-consuming workloads in deep learning. State-of-the-art optimizers, such as AdaGrad, RMSProp and Adam, reduce this labor by adaptively tuning an individual learning rate for each variable. Recently researchers have shown renewed interest in simpler methods like momentum SGD ... | This paper proposes a method to automatically tuning the momentum parameter in momentum SGD methods, which achieves better results and fast convergence speed than state-of-the-art Adam algorithm.
Although the results are promising, I found the presentation of this paper almost inaccessible to me.
First, though a minor ... |
iclr_2018_ByxLBMZCb | With the increasing interest in deeper understanding of the loss surface of many non-convex deep models, this paper presents a unifying framework to study the local/global optima equivalence of the optimization problems arising from training of such non-convex models. Using the local openness property of the underlying... | Summary: The paper focuses on the characterization of the landscape of deep neural networks; i.e., when and why local minima are global, what are the conditions for saddle critical points, etc. The paper covers a somewhat wide range of deep nets (from shallow with linear activation to deeper with non-linear activation)... |
iclr_2018_HyZoi-WRb | Published as a conference paper at ICLR 2018 DEBIASING EVIDENCE APPROXIMATIONS: ON IMPORTANCE-WEIGHTED AUTOENCODERS AND JACKKNIFE VARIATIONAL INFERENCE
The importance-weighted autoencoder (IWAE) approach of Burda et al. (2015) defines a sequence of increasingly tighter bounds on the marginal likelihood of latent variab... | [After author feedback]
I think this is an interesting paper and recommend acceptance. My remaining main comments are described in the response to author feedback below.
[Original review]
The authors introduce jackknife variational inference (JVI), a method for debiasing Monte Carlo objectives such as the importance we... |
iclr_2018_SkBHr1WRW | While existing graph embedding models can generate useful embedding vectors for graph-related tasks, what valuable information can be jointly learned from a graph embedding model is less discussed. In this paper, we consider the possibility of detecting critical structures by a graph embedding model. We propose Ego-CNN... | Dear authors,
Thank you for your contribution to ICLR. The problem you are addressing with your work is important. Your paper is well-motivated. Detecting and exploiting "critical structures" in graphs for graph classification is indeed something that is missing in previous work.
After the introduction you discuss som... |
iclr_2018_rkQkBnJAb | Published as a conference paper at ICLR 2018 IMPROVING GANS USING OPTIMAL TRANSPORT
We present Optimal Transport GAN (OT-GAN), a variant of generative adversarial nets minimizing a new metric measuring the distance between the generator distribution and the data distribution. This metric, which we call mini-batch energ... | There have recently been a set of interesting papers on adapting optimal transport to GANs. This makes a lot of sense. The paper makes some very good connections to the state of the art and those competing approaches. The proposal makes sense from the generative standpoint and it is clear from the paper that the key co... |
iclr_2018_SJlhPMWAW | Deep learning on graphs has become a popular research topic with many applications. However, past work has concentrated on learning graph embedding tasks only, which is in contrast with advances in generative models for images and text. Is it possible to transfer this progress to the domain of graphs? We propose to sid... | This paper studies the problem of learning to generate graphs using deep learning methods. The main challenges of generating graphs as opposed to text or images are said to be the following:
(a) Graphs are discrete structures, and incrementally constructing them would lead to non-differentiability (I don't agree with t... |
iclr_2018_HknbyQbC- | Deep neural networks (DNNs) have been found to be vulnerable to adversarial examples resulting from adding small-magnitude perturbations to inputs. Such adversarial examples can mislead DNNs to produce adversary-selected results. Different attack strategies have been proposed to generate adversarial examples, but how t... | I thank the authors for the thoughtful response and rebuttal. The authors have substantially updated their manuscript and improved the presentation.
Re: Speed. I brought up this point because this was a bulleted item in the Introduction in the earlier version of the manuscript. In the revised manuscript, this bullet po... |
iclr_2018_r1SnX5xCb | Published as a conference paper at ICLR 2018 DEEP SENSING: ACTIVE SENSING USING MULTI- DIRECTIONAL RECURRENT NEURAL NETWORKS
For every prediction we might wish to make, we must decide what to observe (what source of information) and when to observe it. Because making observations is costly, this decision must trade off... | This is a very interesting submission that takes an interesting angle on clinical time series modeling, namely, actively choosing when to measure while simultaneously attempting to impute missing measurements and predict outcomes of interest. The proposed solution formulates everything as a giant learning problem that ... |
iclr_2018_SyJ7ClWCb | Published as a conference paper at ICLR 2018 COUNTERING ADVERSARIAL IMAGES USING INPUT TRANSFORMATIONS
This paper investigates strategies that defend against adversarial-example attacks on image-classification systems by transforming the inputs before feeding them to the system. Specifically, we study applying image tr... | To increase robustness to adversarial attacks, the paper fundamentally proposes to transform an input image before feeding it to a convolutional network classifier. The purpose of the transformation is to erase the high-frequency signals potentially embedded by an adversarial attack.
Strong points:
* To my knowledge, t... |
iclr_2018_HJXyS7bRb | Building chatbots that can accomplish goals such as booking a flight ticket is an unsolved problem in natural language understanding. Much progress has been made to build conversation models using techniques such as sequence2sequence modeling. One challenge in applying such techniques to building goal-oriented conversa... | I like the idea of coupling the language and the conversation model. This is in line with the latest trends of constructing end-to-end NN models that deal with the conversation in a holistic manner. The idea of enforcing information isolation is brilliant. Creating hidden information and allowing the two-party model to... |
iclr_2018_Sy3fJXbA- | While much of the work in the design of convolutional networks over the last five years has revolved around the empirical investigation of the importance of depth, filter sizes, and number of feature channels, recent studies have shown that branching, i.e., splitting the computation along parallel but distinct threads ... | The authors extend the ResNeXt architecture. They substitute the simple add operation with a selection operation for each input in the residual module. The selection of the inputs happens through gate weights, which are sampled at train time. At test time, the gates with the highest values are kept on, while the other ... |
iclr_2018_BkSDMA36Z | A NEW METHOD OF REGION EMBEDDING FOR TEXT CLASSIFICATION
To represent a text as a bag of properly identified "phrases" and use the representation for processing the text is proved to be useful. The key question here is how to identify the phrases and represent them. The traditional method of utilizing n-grams can be re... | The authors present a model for text classification. The parameters of the model are an embedding for each word and a local context unit. The local context unit can be seen as a filter for a convolutional layer, but which filter is used at location i depends on the word at location i (i.e. there is one filter per vocab... |
iclr_2018_BJgd7m0xRZ | Anomaly detection discovers regular patterns in unlabeled data and identifies the non-conforming data points, which in some cases are the result of malicious attacks by adversaries. Learners such as One-Class Support Vector Machines (OCSVMs) have been successfully used in anomaly detection, yet their performance may de... | Although the problem addressed in the paper seems interesting, but there lacks of evidence to support some of the arguments that the authors make. And the paper does not contribute novelty to representation learning, therefore, it is not a good fit for the conference. Detailed critiques are as following:
1. The idea pr... |
iclr_2018_rJ6iJmWCW | In this paper, we propose the generation of accented speech using generative adversarial networks (GANs). Through this work we make two main contributions a) The ability to condition latent representations while generating realistic speech samples b) The ability to efficiently generate long speech samples by using a no... | This paper presents a method for generating speech audio in a particular accent. The proposed approach relies on a generative adversarial network (GAN), combined with a policy approach for joining together generated speech segments. The latter is used to deal with the problem of generating very long sequences (which is... |
iclr_2018_HktRlUlAZ | POLAR TRANSFORMER NETWORKS
Convolutional neural networks (CNNs) are inherently equivariant to translation. Efforts to embed other forms of equivariance have concentrated solely on rotation. We expand the notion of equivariance in CNNs through the Polar Transformer Network (PTN). PTN combines ideas from the Spatial Tran... | This paper presents a new convolutional network architecture that is invariant to global translations and equivariant to rotations and scaling. The method is combination of a spatial transformer module that predicts a focal point, around which a log-polar transform is performed. The resulting log-polar image is analyze... |
iclr_2018_SyqAPeWAZ | In recent years Convolutional Neural Networks (CNN) have been used extensively for Superresolution (SR). In this paper, we use inverse problem and sparse representation solutions to form a mathematical basis for CNN operations. We show how a single neuron is able to provide the optimum solution for inverse problem, giv... | The method proposes a new architecture for solving image super-resolution task. They provide an analysis that connects aims to establish a connection between how CNNs for solving super resolution and solving sparse regularized inverse problems.
The writing of the paper needs improvement. I was not able to understand th... |
iclr_2018_Hy_o3x-0b | There have been multiple attempts with variational auto-encoders (VAE) to learn powerful global representations of complex data using a combination of latent stochastic variables and an autoregressive model over the dimensions of the data. However, for the most challenging natural image tasks the purely autoregressive ... | The paper combines several recent advances on generative modelling including a ladder variational posterior and a PixelCNN decoder together with the proposed convolutional stochastic layers to boost the NLL results of the current VAEs. The numbers in the tables are good but I have several comments on the motivation, or... |
iclr_2018_r1kP7vlRb | Designing a metric manually for unsupervised sequence generation tasks, such as text generation, is essentially difficult. In a such situation, learning a metric of a sequence from data is one possible solution. The previous study, SeqGAN, proposed the framework for unsupervised sequence generation, in which a metric i... | This paper considers the problem of improving sequence generation by learning better metrics. Specifically, it focuses on addressing the exposure bias problem, where traditional methods such as SeqGAN uses GAN framework and reinforcement learning. Different from these work, this paper does not use GAN framework. Instea... |
iclr_2018_Sy8XvGb0- | Published as a conference paper at ICLR 2018 LATENT CONSTRAINTS: LEARNING TO GENERATE CONDITIONALLY FROM UNCONDITIONAL GENERATIVE MODELS
Deep generative neural networks have proven effective at both conditional and unconditional modeling of complex data distributions. Conditional generation enables interactive control,... | UPDATE: I think the authors' rebuttal and updated draft address my points sufficiently well for me to update my score and align myself with the other reviewers.
-----
ORIGINAL REVIEW: The paper proposes a method for learning post-hoc to condition a decoder-based generative model which was trained unconditionally. Start... |
iclr_2018_Skvd-myR- | Measuring visual (dis)similarity between two or more instances within a data distribution is a fundamental task in many applications, especially in image retrieval. Theoretically, non-metric distances are able to generate a more complex and accurate similarity model than metric distances, provided that the non-linear d... | The authors of this work propose learning a similarity measure for visual similarity and obtain, by doing that, an improvement in the very well-known datasets of Oxford and Paris for image retrieval. The work takes high-level image representations generated with an existing architecture (R-MAC), and train on top a neur... |
iclr_2018_H1Nyf7W0Z | Neural sequence generation is commonly approached by using maximumlikelihood (ML) estimation or reinforcement learning (RL). However, it is known that they have their own shortcomings; ML presents training/testing discrepancy, whereas RL suffers from sample inefficiency. We point out that it is difficult to resolve all... | This paper considers a dichitomy between ML and RL based methods for sequence generation. It is argued that the ML approach has some "discrepancy" between the optimization objective and the learning objective, and the RL approach suffers from bad sample complexity. An alpha-divergence formulation is considered to combi... |
iclr_2018_BkpiPMbA- | DECISION BOUNDARY ANALYSIS OF ADVERSARIAL EXAMPLES
Deep neural networks (DNNs) are vulnerable to adversarial examples, which are carefully crafted instances aiming to cause prediction errors for DNNs. Recent research on adversarial examples has examined local neighborhoods in the input space of DNN models. However, pre... | Summary of paper:
The authors present a novel attack for generating adversarial examples, deemed OptMargin, in which the authors attack an ensemble of classifiers created by classifying at random L2 small perturbations. They compare this optimization method with two baselines in MNIST and CIFAR, and provide an analysis... |
iclr_2018_Hksj2WWAW | Published as a conference paper at ICLR 2018 COMBINING SYMBOLIC EXPRESSIONS AND BLACK- BOX FUNCTION EVALUATIONS IN NEURAL PROGRAMS
Neural programming involves training neural networks to learn programs, mathematics, or logic from data. Previous works have failed to achieve good generalization performance, especially on... | Summary
This paper presents a dataset of mathematical equations and applies TreeLSTMs to two tasks: verifying and completing mathematical equations. For these tasks, TreeLSTMs outperform TreeNNs and RNNs. In my opinion, the main contribution of this paper is this potentially useful dataset, as well as an interesting wa... |
iclr_2018_HJhIM0xAW | Published as a conference paper at ICLR 2018 LEARNING A NEURAL RESPONSE METRIC FOR RETINAL PROSTHESIS
Retinal prostheses for treating incurable blindness are designed to electrically stimulate surviving retinal neurons, causing them to send artificial visual signals to the brain. However, electrical stimulation general... | * Summary of paper: The paper addresses the problem of optimizing metrics in the context of retinal prosthetics: Their goal is to learn a metric which assumes spike-patterns generated by the same stimulus to be more similar to each other than spike-patterns generated by different stimuli. They compare a conventional, q... |
iclr_2018_SyGT_6yCZ | The quality of the features used in visual recognition is of fundamental importance for the overall system. For a long time, low-level hand-designed feature algorithms as SIFT and HOG have obtained the best results on image recognition. Visual features have recently been extracted from trained convolutional neural netw... | This paper deals with early stopping but the contributions are limited. This work would fit better a workshop as a preliminary result, furthermore it is too short. Following a short review section per section.
Intro: The name SFC is misleading as the method consists in stopping early the training with an optimized lear... |
iclr_2018_SkqV-XZRZ | Recurrent neural networks like long short-term memory (LSTM) are important architectures for sequential prediction tasks. LSTMs (and RNNs in general) model sequences along the forward time direction. Bidirectional LSTMs (Bi-LSTMs), which model sequences along both forward and backward directions, generally perform bett... | *Quality*
The paper is easy to parse, with clear diagrams and derivations at the start. The problem context is clearly stated, as is the proposed model.
The improvements in terms of average log-likelihood are clear. The model does improve over state-of-the-art in some cases, but not all.
Based on the presented findings... |
iclr_2018_S1XXq6lRW | Labeled text classification datasets are typically only available in a few select languages. In order to train a model for e.g news categorization in a language L t without a suitable text classification dataset there are two options. The first option is to create a new labeled dataset by hand, and the second option is... | This paper addresses the problem of learning a cross-language text categorizer with no labelled information in the target language. The suggested solution relies on learning cross-lingual embeddings, and training a classifier using labelled data in the source language only.
The idea of using cross-lingual or multilingu... |
iclr_2018_HJGv1Z-AW | EMERGENCE OF LINGUISTIC COMMUNICATION FROM REFERENTIAL GAMES WITH SYMBOLIC AND PIXEL INPUT
The ability of algorithms to evolve or learn (compositional) communication protocols has traditionally been studied in the language evolution literature through the use of emergent communication tasks. Here we scale up this resea... | --------------
Summary:
--------------
This paper presents a series of experiments on language emergence through referential games between two agents. They ground these experiments in both fully-specified symbolic worlds and through raw, entangled, visual observations of simple synthetic scenes. They provide rich analy... |
iclr_2018_SJa1Nk10b | Workshop track -ICLR 2018 ANYTIME NEURAL NETWORK: A VERSATILE TRADE- OFF BETWEEN COMPUTATION AND ACCURACY
We present an approach for anytime predictions in deep neural networks (DNNs). For each test sample, an anytime predictor produces a coarse result quickly, and then continues to refine it until the test-time comput... | 1. Paper Summary
This paper adds a separate network at every layer of a residual network that performs classification. They minimize the loss of every classifier using two proposed weighting schemes. They also ensemble this model.
2. High level paper
The organization of this paper is a bit confusing. Two weighing schem... |
iclr_2018_Sy21R9JAW | TOWARDS BETTER UNDERSTANDING OF GRADIENT-BASED ATTRIBUTION METHODS FOR DEEP NEURAL NETWORKS
Understanding the flow of information in Deep Neural Networks (DNNs) is a challenging problem that has gain increasing attention over the last few years. While several methods have been proposed to explain network predictions, t... | The paper summarizes and compares some of the current explanation techniques for deep neural networks that rely on the redistribution of relevance / contribution values from the output to the input space.
The main contributions are the introduction of a unified framework that expresses 4 common attribution techniques (... |
iclr_2018_B16_iGWCW | In this paper, a deep boosting algorithm is developed to learn more discriminative ensemble classifier by seamlessly combining a set of base deep CNNs (base experts) with diverse capabilities, e.g., these base deep CNNs are sequentially trained to recognize a set of object classes in an easy-to-hard way according to th... | This paper consider a version of boosting where in each iteration only class weights are updated rather than sample weights and apply that to a series of CNNs for object recognition tasks.
While the paper is comprehensive in their derivations (very similar to original boosting papers and in many cases one to one transl... |
iclr_2018_SyjsLqxR- | Classifiers such as deep neural networks have been shown to be vulnerable against adversarial perturbations on problems with high-dimensional input space. While adversarial training improves the robustness of classifiers against such adversarial perturbations, it leaves classifiers sensitive to them on a non-negligible... | Summary:
This paper empirically studies adversarial perturbations dx and what the effects are of adversarial training (AT) with respect to shared (dx fools for many x) and singular (only for a single x) perturbations. Experiments use a (previously published) iterative fast-gradient-sign-method and use a Resnet on CIFAR... |
iclr_2018_rk3mjYRp- | Policy gradients methods often achieve better performance when the change in policy is limited to a small Kullback-Leibler divergence. We derive policy gradients where the change in policy is limited to a small Wasserstein distance (or trust region). This is done in the discrete and continuous multi-armed bandit settin... | The paper ‘Diffusing policies: Towards Wasserstein policy gradient flows’ explores
the connections between reinforcement learning and the theory of quadratic optimal transport (i.e.
using the Wasserstein_2 as a regularizer of an iterative problem that converges toward
an optimal policy). Following a classical result f... |
iclr_2018_Sy2ogebAW | Published as a conference paper at ICLR 2018 UNSUPERVISED NEURAL MACHINE TRANSLATION
In spite of the recent success of neural machine translation (NMT) in standard benchmarks, the lack of large parallel corpora poses a major practical problem for many language pairs. There have been several proposals to alleviate this ... | unsupervised neural machine translation
This is an interesting paper on unsupervised MT. It trains a standard architecture using:
1) word embeddings in a shared embedding space, learned using a recent approach that works with only tens of bilingual word papers.
2) A encoder-decoder trained using only monolingual data (... |
iclr_2018_BJJLHbb0- | Published as a conference paper at ICLR 2018 DEEP AUTOENCODING GAUSSIAN MIXTURE MODEL FOR UNSUPERVISED ANOMALY DETECTION
Unsupervised anomaly detection on multi-or high-dimensional data is of great importance in both fundamental machine learning research and industrial applications, for which density estimation lies at... | The paper presents a new technique for anomaly detection where the dimension reduction and the density estimation steps are jointly optimized. The paper is rigorous and ideas are clearly stated. The idea to constraint the dimension reduction to fit a certain model, here a GMM, is relevant, and the paper provides a thor... |
iclr_2018_rkTS8lZAb | Published as a conference paper at ICLR 2018 BOUNDARY-SEEKING GENERATIVE ADVERSARIAL NETWORKS
Generative adversarial networks (GANs, Goodfellow et al., 2014) are a learning framework that rely on training a discriminator to estimate a measure of difference between a target and generated distributions. GANs, as normally... | Thanks for the feedback and for clarifying the 1) algorithm and the assumptions in the multivariate case 2) comparison to RL based methods 3) connection to estimating importance sampling weights using GAN discriminator.
I think the paper contribution is now more clear and strengthened with additional convincing experim... |
iclr_2018_BJk59JZ0b | GUIDE ACTOR-CRITIC FOR CONTINUOUS CONTROL
Actor-critic methods solve reinforcement learning problems by updating a parameterized policy known as an actor in a direction that increases an estimate of the expected return known as a critic. However, existing actor-critic methods only use values or gradients of the critic ... | The paper presents a clever trick for updating the actor in an actor-critic setting: computing a guide actor that diverges from the actor to improve critic value, then updating the actor parameters towards the guide actor. This can be done since, when the parametrized actor is Gaussian and the critic value can be well-... |
iclr_2018_ryDNZZZAW | While domain adaptation has been actively researched in recent years, most theoretical results and algorithms focus on the single-source-single-target adaptation setting. Naive application of such algorithms on multiple source domain adaptation problem may lead to suboptimal solutions. We propose a new generalization b... | Quality:
The paper appears to be correct.
Clarity:
The paper is very clear
Originality:
The theoretical contribution extends the seminal work of Ben-David et al., the idea of using adversarial learning is not new, the novelty is mediaum
Significance:
The theoretical analysis is interested but for me limited, the idea o... |
iclr_2018_rJWechg0Z | MINIMAL-ENTROPY CORRELATION ALIGNMENT FOR UNSUPERVISED DEEP DOMAIN ADAPTATION
In this work, we face the problem of unsupervised domain adaptation with a novel deep learning approach which leverages our finding that entropy minimization is induced by the optimal alignment of second order statistics between source and ta... | This paper improves the correlation alignment approach to domain adaptation from two aspects. One is to replace the Euclidean distance by the geodesic Log-Euclidean distance between two covariance matrices. The other is to automatically select the balancing cost by the entropy on the target domain. Experiments are cond... |
iclr_2018_r1TA9ZbA- | Planning problems are among the most important and well-studied problems in artificial intelligence. They are most typically solved by tree search algorithms that simulate ahead into the future, evaluate future states, and back-up those evaluations to the root of a search tree. Among these algorithms, Monte-Carlo tree ... | This paper designs a deep learning architecture that mimics the structure of the well-known MCTS algorithm. From gold standard state-action pairs, it learns each component of this architecture in order to predict similar actions.
I enjoyed reading this paper. The presentation is very clear, the design of the architectu... |
iclr_2018_HJWLfGWRb | Published as a conference paper at ICLR 2018 MATRIX CAPSULES WITH EM ROUTING
A capsule is a group of neurons whose outputs represent different properties of the same entity. Each layer in a capsule network contains many capsules. We describe a version of capsules in which each capsule has a logistic unit to represent t... | The paper proposes a novel architecture for capsule networks. Each capsule has a logistic unit representing the presence of an entity plus a 4x4 pose matrix representing the entity/viewer relationship. This new representation comes with a novel iterative routing scheme, based on the EM algorithm.
Evaluated on the Small... |
iclr_2018_HkepKG-Rb | This paper develops a novel methodology for using symbolic knowledge in deep learning. From first principles, we derive a semantic loss function that bridges between neural output vectors and logical constraints. This loss function captures how close the neural network is to satisfying the constraints on its output. An... | SUMMARY
The paper proposes a new form of regularization utilizing logical constraints. The semantic loss function is built on the exploitation of symbolic knowledge extracted from data and connecting the logical constraints to the outputs of a neural network. The use of Boolean logic as a constraint provides a seconda... |
iclr_2018_SkOb1Fl0Z | Workshop track -ICLR 2018 A FLEXIBLE APPROACH TO AUTOMATED RNN ARCHITECTURE GENERATION
The process of designing neural architectures requires expert knowledge and extensive trial and error. While automated architecture search may simplify these requirements, the recurrent neural network (RNN) architectures generated by... | This paper investigates meta-learning strategy for automated architecture search in the context of RNN. To constraint the architecture search space, authors propose a DSL that specifies the RNN recurrent operations. This DSL allows to explore RNN architectures using either random search or a reinforcement-learning stra... |
iclr_2018_HJWGdbbCW | We propose a general deep reinforcement learning method and apply it to robot manipulation tasks. Our approach leverages demonstration data to assist a reinforcement learning agent in learning to solve a wide range of tasks, mainly previously unsolved. We train visuomotor policies end-to-end to learn a direct mapping f... | Paper summary: The authors propose a number of tricks to enable training policies for pick and place style tasks using a combination of GAIL-based imitation learning and hand-specified rewards, as well as use of unobserved state information during training and hand-designed curricula. The results demonstrate manipulati... |
iclr_2018_BJj6qGbRW | FEW-SHOT LEARNING WITH GRAPH NEURAL NET- WORKS
We propose to study the problem of few-shot learning with the prism of inference on a partially observed graphical model, constructed from a collection of input images whose label can be either observed or not. By assimilating generic message-passing inference algorithms w... | This paper introduces a graph neural net approach to few-shot learning. Input examples form the nodes of the graph and edge weights are computed as a nonlinear function of the absolute difference between node features. In addition to standard supervised few-shot classification, both semi-supervised and active learning ... |
iclr_2018_BkVsWbbAW | Despite advances in deep learning, artificial neural networks do not learn the same way as humans do. Today, neural networks can learn multiple tasks when trained on them jointly, but cannot maintain performance on learnt tasks when tasks are presented one at a time -this phenomenon called catastrophic forgetting is a ... | This paper introduces a neural network architecture for continual learning. The model is inspired by current knowledge about long term memory consolidation mechanisms in humans. As a consequence, it uses:
- One temporary memory storage (inspired by hippocampus) and a long term memory
- A notion of memory replay, implem... |
iclr_2018_r1gs9JgRZ | Published as a conference paper at ICLR 2018 MIXED PRECISION TRAINING
Increasing the size of a neural network typically improves accuracy but also increases the memory and compute requirements for training the model. We introduce methodology for training deep neural networks using half-precision floating point numbers,... | The paper considers the problem of training neural networks in mixed precision (MP), using both 16-bit floating point (FP16) and 32-bit floating point (FP32). The paper proposes three techniques for training networks in mixed precision: first, keep a master copy of network parameters in FP32; second, use loss scaling t... |
iclr_2018_r1Kr3TyAb | We conduct mathematical analysis on the effect of batch normalization (BN) on gradient backpropogation in residual network training, which is believed to play a critical role in addressing the gradient vanishing/explosion problem, in this work. By analyzing the mean and variance behavior of the input and the gradient i... | This paper attempts to analyze the gradient flow through a batchNorm-ReLU ResNet and make suggestions for reducing gradient explosion.
Firstly, the paper has a fatal mathematical flaw. Consider equation (10). There, you show the variance of y_{L,i} taken over BOTH random weights AND the batch. Now consider equation (32... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.