Paper Group NANR 62
Why Does the VQA Model Answer No?: Improving Reasoning through Visual and Linguistic Inference. Auto Network Compression with Cross-Validation Gradient. Regularizing activations in neural networks via distribution matching with the Wassertein metric. TrojanNet: Exposing the Danger of Trojan Horse Attack on Neural Networks. Sparse Coding with Gated …
Why Does the VQA Model Answer No?: Improving Reasoning through Visual and Linguistic Inference
| Title | Why Does the VQA Model Answer No?: Improving Reasoning through Visual and Linguistic Inference |
| Authors | Anonymous |
| Abstract | In order to make Visual Question Answering (VQA) explainable, previous studies not only visualize the attended region of a VQA model but also generate textual explanations for its answers. However, when the model’s answer is ‘no,’ existing methods have difficulty in revealing detailed arguments that lead to that answer. In addition, previous methods are insufficient to provide logical bases, when the question requires common sense to answer. In this paper, we propose a novel textual explanation method to overcome the aforementioned limitations. First, we extract keywords that are essential to infer an answer from a question. Second, for a pre-trained explanation generator, we utilize a novel Variable-Constrained Beam Search (VCBS) algorithm to generate phrases that best describes the relationship between keywords in images. Then, we complete an explanation by feeding the phrase to the generator. Furthermore, if the answer to the question is “yes” or “no,” we apply Natural Langauge Inference (NLI) to identify whether contents of the question can be inferred from the explanation using common sense. Our user study, conducted in Amazon Mechanical Turk (MTurk), shows that our proposed method generates more reliable explanations compared to the previous methods. Moreover, by modifying the VQA model’s answer through the output of the NLI model, we show that VQA performance increases by 1.1% from the original model. |
| Tasks | Common Sense Reasoning, Question Answering, Visual Question Answering |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HJlvCR4KDS |
| https://openreview.net/pdf?id=HJlvCR4KDS | |
| PWC | https://paperswithcode.com/paper/why-does-the-vqa-model-answer-no-improving |
| Repo | |
| Framework | |
Auto Network Compression with Cross-Validation Gradient
| Title | Auto Network Compression with Cross-Validation Gradient |
| Authors | Anonymous |
| Abstract | Network compression technology can compress large and complex networks into small networks, so that it can be deployed on devices with limited resources. Sparse regularization method, such as $\normlone$ or $L^{21}$ regularization, is the most popular method that can induce sparse model. However, it introduces new hyperparameters, which not only affects the degree of sparsity, but also involves whether the network can be effectively trained (gradient explosion or model non-convergence). How to select hyperparameters becomes an important and open problem for regularization-based network compression method. In this paper, we propose an auto network compression framework with cross-validation gradient which can automatically adjust the hyperparameters. Firstly, we design an unified framework which combines model parameter learning with hyperparametric learning. Secondly, in order to solve the problem of non-derivability of $\normlone$ norm, we introduce auxiliary variables to transform it into a solvable problem, and then obtain the derivative of model parameters with respect to hyperparameters. Finally, the derivative of the hyperparametric vector is solved by the chain rule. In solving the inverse problem of Heisen matrix, we compare three methods and only calculate the mixed partial derivatives. To a certain extent, this method realizes the automatic network compression. Classical network structures such as VGG, ResNet and DensNet are tested on CIFAR-10 and CIFAR-100 datasets to prove the effectiveness of our algorithm. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=r1eoflSFvS |
| https://openreview.net/pdf?id=r1eoflSFvS | |
| PWC | https://paperswithcode.com/paper/auto-network-compression-with-cross |
| Repo | |
| Framework | |
Regularizing activations in neural networks via distribution matching with the Wassertein metric
| Title | Regularizing activations in neural networks via distribution matching with the Wassertein metric |
| Authors | Anonymous |
| Abstract | Regularization and normalization have become an indispensable component in deep learning because it enables faster training and improved generalization performance. We propose the projected error function regularization loss (PER) that encourages activations to follow the standard normal distribution. PER randomly projects activations to one dimensional space and computes the regularization in the projected space. PER acts like the Pseudo-Huber loss in the projected space, enabling robust regularization for training deep neural networks. In addition, PER can capture interaction between hidden units by projection vector drawn from unit sphere. By doing so, PER minimizes the upper bound of the Wasserstein distance of order one between an empirical distribution of activations and the standard normal distribution. To the best of the authors’ knowledge, this is the first work to regularize activations concerning the target distribution in the probability distribution space. We evaluate the proposed method on image classification task and word-level language modeling task. |
| Tasks | Image Classification, Language Modelling |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rygwLgrYPB |
| https://openreview.net/pdf?id=rygwLgrYPB | |
| PWC | https://paperswithcode.com/paper/regularizing-activations-in-neural-networks |
| Repo | |
| Framework | |
TrojanNet: Exposing the Danger of Trojan Horse Attack on Neural Networks
| Title | TrojanNet: Exposing the Danger of Trojan Horse Attack on Neural Networks |
| Authors | Chuan Guo, Ruihan Wu, Kilian Q. Weinberger |
| Abstract | The complexity of large-scale neural networks can lead to poor understanding of their internal details. We show that this opaqueness provides an opportunity for adversaries to embed unintended functionalities into the network in the form of Trojan horse attacks. Our novel framework hides the existence of a malicious network within a benign transport network. Our attack is flexible, easy to execute, and difficult to detect. We prove theoretically that the malicious network’s detection is computationally infeasible and demonstrate empirically that the transport network does not compromise its disguise. Our attack exposes an important, previously unknown loophole that unveils a new direction in machine learning security. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BJeGA6VtPS |
| https://openreview.net/pdf?id=BJeGA6VtPS | |
| PWC | https://paperswithcode.com/paper/trojannet-exposing-the-danger-of-trojan-horse |
| Repo | |
| Framework | |
Sparse Coding with Gated Learned ISTA
| Title | Sparse Coding with Gated Learned ISTA |
| Authors | Anonymous |
| Abstract | In this paper, we study the learned iterative shrinkage thresholding algorithm (LISTA) for solving sparse coding problems. Following assumptions made by prior works, we first discover that the code components in its estimations may be lower than expected, i.e., require gains, and to address this problem, a gated mechanism amenable to theoretical analysis is then introduced. Specific design of the gates is inspired by convergence analyses of the mechanism and hence its effectiveness can be formally guaranteed. In addition to the gain gates, we further introduce overshoot gates for compensating insufficient step size in LISTA. Extensive empirical results confirm our theoretical findings and verify the effectiveness of our method. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BygPO2VKPH |
| https://openreview.net/pdf?id=BygPO2VKPH | |
| PWC | https://paperswithcode.com/paper/sparse-coding-with-gated-learned-ista |
| Repo | |
| Framework | |
Making the Shoe Fit: Architectures, Initializations, and Tuning for Learning with Privacy
| Title | Making the Shoe Fit: Architectures, Initializations, and Tuning for Learning with Privacy |
| Authors | Anonymous |
| Abstract | Because learning sometimes involves sensitive data, standard machine-learning algorithms have been extended to offer strong privacy guarantees for training data. However, in practice, this has been mostly an afterthought, with privacy-preserving models obtained by re-running training with a different optimizer, but using the same model architecture that performed well in a non-privacy-preserving setting. This approach leads to less than ideal privacy/utility tradeoffs, as we show here. Instead, we propose that model architectures and initializations are chosen and hyperparameter tuning is performed, ab initio, explicitly for privacy-preserving training. Using this paradigm, we achieve new state-of-the-art accuracy on MNIST, FashionMNIST, and CIFAR10 without any modification of the fundamental learning procedures or differential-privacy analysis. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJg851rYwH |
| https://openreview.net/pdf?id=rJg851rYwH | |
| PWC | https://paperswithcode.com/paper/making-the-shoe-fit-architectures |
| Repo | |
| Framework | |
Model-free Learning Control of Nonlinear Stochastic Systems with Stability Guarantee
| Title | Model-free Learning Control of Nonlinear Stochastic Systems with Stability Guarantee |
| Authors | Anonymous |
| Abstract | Reinforcement learning (RL) offers a principled way to achieve the optimal cumulative performance index in discrete-time nonlinear stochastic systems, which are modeled as Markov decision processes. Its integration with deep learning techniques has promoted the field of deep RL with an impressive performance in complicated continuous control tasks. However, from a control-theoretic perspective, the first and most important property of a system to be guaranteed is stability. Unfortunately, stability is rarely assured in RL and remains an open question. In this paper, we propose a stability guaranteed RL framework which simultaneously learns a Lyapunov function along with the controller or policy, both of which are parameterized by deep neural networks, by borrowing the concept of Lyapunov function from control theory. Our framework can not only offer comparable or superior control performance over state-of-the-art RL algorithms, but also construct a Lyapunov function to validate the closed-loop stability. In the simulated experiments, our approach is evaluated on several well-known examples including classic CartPole balancing, 3-dimensional robot control and control of synthetic biology gene regulatory networks. Compared with RL algorithms without stability guarantee, our approach can enable the system to recover to the operating point when interfered by uncertainties such as unseen disturbances and system parametric variations to a certain extent. |
| Tasks | Continuous Control |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SkeWc2EKPH |
| https://openreview.net/pdf?id=SkeWc2EKPH | |
| PWC | https://paperswithcode.com/paper/model-free-learning-control-of-nonlinear |
| Repo | |
| Framework | |
Scoring-Aggregating-Planning: Learning task-agnostic priors from interactions and sparse rewards for zero-shot generalization
| Title | Scoring-Aggregating-Planning: Learning task-agnostic priors from interactions and sparse rewards for zero-shot generalization |
| Authors | Anonymous |
| Abstract | Humans can learn task-agnostic priors from interactive experience and utilize the priors for novel tasks without any finetuning. In this paper, we propose Scoring-Aggregating-Planning (SAP), a framework that can learn task-agnostic semantics and dynamics priors from arbitrary quality interactions as well as the corresponding sparse rewards and then plan on unseen tasks in zero-shot condition. The framework finds a neural score function for local regional state and action pairs that can be aggregated to approximate the quality of a full trajectory; moreover, a dynamics model that is learned with self-supervision can be incorporated for planning. Many of previous works that leverage interactive data for policy learning either need massive on-policy environmental interactions or assume access to expert data while we can achieve a similar goal with pure off-policy imperfect data. Instantiating our framework results in a generalizable policy to unseen tasks. Experiments demonstrate that the proposed method can outperform baseline methods on a wide range of applications including gridworld, robotics tasks and video games. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HyeuP2EtDB |
| https://openreview.net/pdf?id=HyeuP2EtDB | |
| PWC | https://paperswithcode.com/paper/scoring-aggregating-planning-learning-task-2 |
| Repo | |
| Framework | |
A Mean-Field Theory for Kernel Alignment with Random Features in Generative Adverserial Networks
| Title | A Mean-Field Theory for Kernel Alignment with Random Features in Generative Adverserial Networks |
| Authors | Anonymous |
| Abstract | We propose a novel supervised learning method to optimize the kernel in maximum mean discrepancy generative adversarial networks (MMD GANs). Specifically, we characterize a distributionally robust optimization problem to compute a good distribution for the random feature model of Rahimi and Recht to approximate a good kernel function. Due to the fact that the distributional optimization is infinite dimensional, we consider a Monte-Carlo sample average approximation (SAA) to obtain a more tractable finite dimensional optimization problem. We subsequently leverage a particle stochastic gradient descent (SGD) method to solve finite dimensional optimization problems. Based on a mean-field analysis, we then prove that the empirical distribution of the interactive particles system at each iteration of the SGD follows the path of the gradient descent flow on the Wasserstein manifold. We also establish the non-asymptotic consistency of the finite sample estimator. Our empirical evaluation on synthetic data-set as well as MNIST and CIFAR-10 benchmark data-sets indicates that our proposed MMD GAN model with kernel learning indeed attains higher inception scores well as Fr`{e}chet inception distances and generates better images compared to the generative moment matching network (GMMN) and MMD GAN with untrained kernels. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Skl-fyHKPH |
| https://openreview.net/pdf?id=Skl-fyHKPH | |
| PWC | https://paperswithcode.com/paper/a-mean-field-theory-for-kernel-alignment-with-1 |
| Repo | |
| Framework | |
Generative Ratio Matching Networks
| Title | Generative Ratio Matching Networks |
| Authors | Anonymous |
| Abstract | Deep generative models can learn to generate realistic-looking images, but many of the most effective methods are adversarial and involve a saddlepoint optimization, which require careful balancing of training between a generator network and a critic network. Maximum mean discrepancy networks (MMD-nets) avoid this issue by using kernel as a fixed adversary, but unfortunately they have not on their own been able to match the generative quality of adversarial training. In this work, we take their insight of using kernels as fixed adversaries further and present a novel method for training deep generative models that does not involve saddlepoint optimization. We call our method generative ratio matching or GRAM for short. In GRAM, the generator and the critic networks do not play a zero-sum game against each other, instead they do so against a fixed kernel. Thus GRAM networks are not only stable to train like MMD-nets but they also match and beat the generative quality of adversarially trained generative networks. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SJg7spEYDS |
| https://openreview.net/pdf?id=SJg7spEYDS | |
| PWC | https://paperswithcode.com/paper/generative-ratio-matching-networks |
| Repo | |
| Framework | |
Event Discovery for History Representation in Reinforcement Learning
| Title | Event Discovery for History Representation in Reinforcement Learning |
| Authors | Anonymous |
| Abstract | Environments in Reinforcement Learning (RL) are usually only partially observable. To address this problem, a possible solution is to provide the agent with information about past observations. While common methods represent this history using a Recurrent Neural Network (RNN), in this paper we propose an alternative representation which is based on the record of the past events observed in a given episode. Inspired by the human memory, these events describe only important changes in the environment and, in our approach, are automatically discovered using self-supervision. We evaluate our history representation method using two challenging RL benchmarks: some games of the Atari-57 suite and the 3D environment Obstacle Tower. Using these benchmarks we show the advantage of our solution with respect to common RNN-based approaches. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=H1eVlgHKPr |
| https://openreview.net/pdf?id=H1eVlgHKPr | |
| PWC | https://paperswithcode.com/paper/event-discovery-for-history-representation-in |
| Repo | |
| Framework | |
Low-Resource Knowledge-Grounded Dialogue Generation
| Title | Low-Resource Knowledge-Grounded Dialogue Generation |
| Authors | Anonymous |
| Abstract | Responding with knowledge has been recognized as an important capability for an intelligent conversational agent. Yet knowledge-grounded dialogues, as training data for learning such a response generation model, are difficult to obtain. Motivated by the challenge in practice, we consider knowledge-grounded dialogue generation under a natural assumption that only limited training examples are available. In such a low-resource setting, we devise a disentangled response decoder in order to isolate parameters that depend on knowledge-grounded dialogues from the entire generation model. By this means, the major part of the model can be learned from a large number of ungrounded dialogues and unstructured documents, while the remaining small parameters can be well fitted using the limited training examples. Evaluation results on two benchmarks indicate that with only $1/8$ training data, our model can achieve the state-of-the-art performance and generalize well on out-of-domain knowledge. |
| Tasks | Dialogue Generation |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJeIcTNtvS |
| https://openreview.net/pdf?id=rJeIcTNtvS | |
| PWC | https://paperswithcode.com/paper/low-resource-knowledge-grounded-dialogue |
| Repo | |
| Framework | |
Semi-Supervised Generative Modeling for Controllable Speech Synthesis
| Title | Semi-Supervised Generative Modeling for Controllable Speech Synthesis |
| Authors | Anonymous |
| Abstract | We present a novel generative model that combines state-of-the-art neural text- to-speech (TTS) with semi-supervised probabilistic latent variable models. By providing partial supervision to some of the latent variables, we are able to force them to take on consistent and interpretable purposes, which previously hasn’t been possible with purely unsupervised methods. We demonstrate that our model is able to reliably discover and control important but rarely labelled attributes of speech, such as affect and speaking rate, with as little as 1% (30 minutes) supervision. Even at such low supervision levels we do not observe a degradation of synthesis quality compared to a state-of-the-art baseline. We will release audio samples at https://tts-demos.github.io/. |
| Tasks | Latent Variable Models, Speech Synthesis |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJeqeCEtvH |
| https://openreview.net/pdf?id=rJeqeCEtvH | |
| PWC | https://paperswithcode.com/paper/semi-supervised-generative-modeling-for-1 |
| Repo | |
| Framework | |
BOSH: An Efficient Meta Algorithm for Decision-based Attacks
| Title | BOSH: An Efficient Meta Algorithm for Decision-based Attacks |
| Authors | Anonymous |
| Abstract | Adversarial example generation becomes a viable method for evaluating the robustness of a machine learning model. In this paper, we consider hard-label black- box attacks (a.k.a. decision-based attacks), which is a challenging setting that generates adversarial examples based on only a series of black-box hard-label queries. This type of attacks can be used to attack discrete and complex models, such as Gradient Boosting Decision Tree (GBDT) and detection-based defense models. Existing decision-based attacks based on iterative local updates often get stuck in a local minimum and fail to generate the optimal adversarial example with the smallest distortion. To remedy this issue, we propose an efficient meta algorithm called BOSH-attack, which tremendously improves existing algorithms through Bayesian Optimization (BO) and Successive Halving (SH). In particular, instead of traversing a single solution path when searching an adversarial example, we maintain a pool of solution paths to explore important regions. We show empirically that the proposed algorithm converges to a better solution than existing approaches, while the query count is smaller than applying multiple random initializations by a factor of 10. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=ryxPbkrtvr |
| https://openreview.net/pdf?id=ryxPbkrtvr | |
| PWC | https://paperswithcode.com/paper/bosh-an-efficient-meta-algorithm-for-decision |
| Repo | |
| Framework | |
Feature Selection using Stochastic Gates
| Title | Feature Selection using Stochastic Gates |
| Authors | Anonymous |
| Abstract | Feature selection problems have been extensively studied in the setting of linear estimation, for instance LASSO, but less emphasis has been placed on feature selection for non-linear functions. In this study, we propose a method for feature selection in high-dimensional non-linear function estimation problems. The new procedure is based on directly penalizing the $\ell_0$ norm of features, or the count of the number of selected features. Our $\ell_0$ based regularization relies on a continuous relaxation of the Bernoulli distribution, which allows our model to learn the parameters of the approximate Bernoulli distributions via gradient descent. The proposed framework simultaneously learns a non-linear regression or classification function while selecting a small subset of features. We provide an information-theoretic justification for incorporating Bernoulli distribution into our approach. Furthermore, we evaluate our method using synthetic and real-life data and demonstrate that our approach outperforms other embedded methods in terms of predictive performance and feature selection. |
| Tasks | Feature Selection |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=B1lda1HtvB |
| https://openreview.net/pdf?id=B1lda1HtvB | |
| PWC | https://paperswithcode.com/paper/feature-selection-using-stochastic-gates-1 |
| Repo | |
| Framework | |
