Paper Group NANR 19
Star-Convexity in Non-Negative Matrix Factorization. Domain Adaptive Multiflow Networks. Connectivity-constrained interactive annotations for panoptic segmentation. Learning to Move with Affordance Maps. AdvCodec: Towards A Unified Framework for Adversarial Text Generation. Multi-task Network Embedding with Adaptive Loss Weighting. OPTIMAL TRANSPOR …
Star-Convexity in Non-Negative Matrix Factorization
| Title | Star-Convexity in Non-Negative Matrix Factorization |
| Authors | Anonymous |
| Abstract | Non-negative matrix factorization (NMF) is a highly celebrated algorithm for matrix decomposition that guarantees strictly non-negative factors. The underlying optimization problem is computationally intractable, yet in practice gradient descent based solvers often find good solutions. This gap between computational hardness and practical success mirrors recent observations in deep learning, where it has been the focus of extensive discussion and analysis. In this paper we revisit the NMF optimization problem and analyze its loss landscape in non-worst-case settings. It has recently been observed that gradients in deep networks tend to point towards the final minimizer throughout the optimization. We show that a similar property holds (with high probability) for NMF, provably in a non-worst case model with a planted solution, and empirically across an extensive suite of real-world NMF problems. Our analysis predicts that this property becomes more likely with growing number of parameters, and experiments suggest that a similar trend might also hold for deep neural networks — turning increasing data sets and models into a blessing from an optimization perspective. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BylKwnEYvS |
| https://openreview.net/pdf?id=BylKwnEYvS | |
| PWC | https://paperswithcode.com/paper/star-convexity-in-non-negative-matrix |
| Repo | |
| Framework | |
Domain Adaptive Multiflow Networks
| Title | Domain Adaptive Multiflow Networks |
| Authors | Anonymous |
| Abstract | We tackle unsupervised domain adaptation by accounting for the fact that different domains may need to be processed differently to arrive to a common feature representation effective for recognition. To this end, we introduce a deep learning framework where each domain undergoes a different sequence of operations, allowing some, possibly more complex, domains to go through more computations than others. This contrasts with state-of-the-art domain adaptation techniques that force all domains to be processed with the same series of operations, even when using multi-stream architectures whose parameters are not shared. As evidenced by our experiments, the greater flexibility of our method translates to higher accuracy. Furthermore, it allows us to handle any number of domains simultaneously. |
| Tasks | Domain Adaptation, Unsupervised Domain Adaptation |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJxycxHKDS |
| https://openreview.net/pdf?id=rJxycxHKDS | |
| PWC | https://paperswithcode.com/paper/domain-adaptive-multiflow-networks |
| Repo | |
| Framework | |
Connectivity-constrained interactive annotations for panoptic segmentation
| Title | Connectivity-constrained interactive annotations for panoptic segmentation |
| Authors | Anonymous |
| Abstract | Large-scale ground truth data sets are of crucial importance for deep learning based segmentation models, but annotating per-pixel masks is prohibitively time consuming. In this paper, we investigate interactive graph-based segmentation algorithms that enforce connectivity. To be more precise, we introduce an instance-aware heuristic of a discrete Potts model, and a class-aware Integer Linear Programming (ILP) formulation that ensures global optimum. Both algorithms can take RGB, or utilize the feature maps from any DCNN, whether trained on the target dataset or not, as input. We present competitive semantic (and panoptic) segmentation results on the PASCAL VOC 2012 and Cityscapes dataset given initial scribbles. We also demonstrate that our interactive approach can reach $90.6%$ mIoU on VOC validation set with an overhead of just $3$ correction scribbles. They are thus suitable for interactive annotation on new or existing datasets, or can be used inside any weakly supervised learning framework on new datasets. |
| Tasks | Panoptic Segmentation |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HkliveStvH |
| https://openreview.net/pdf?id=HkliveStvH | |
| PWC | https://paperswithcode.com/paper/connectivity-constrained-interactive |
| Repo | |
| Framework | |
Learning to Move with Affordance Maps
| Title | Learning to Move with Affordance Maps |
| Authors | Anonymous |
| Abstract | The ability to autonomously explore and navigate a physical space is a fundamental requirement for virtually any mobile autonomous agent, from household robotic vacuums to autonomous vehicles. Traditional SLAM-based approaches for exploration and navigation largely focus on leveraging scene geometry, but fail to model dynamic objects (such as other agents) or semantic constraints (such as wet floors or doorways). Learning-based RL agents are an attractive alternative because they can incorporate both semantic and geometric information, but are notoriously sample inefficient, difficult to generalize to novel settings, and are difficult to interpret. In this paper, we combine the best of both worlds with a modular approach that {\em learns} a spatial representation of a scene that is trained to be effective when coupled with traditional geometric planners. Specifically, we design an agent that learns to predict a spatial affordance map that elucidates what parts of a scene are navigable through active self-supervised experience gathering. In contrast to most simulation environments that assume a static world, we evaluate our approach in the VizDoom simulator, using large-scale randomly-generated maps containing a variety of dynamic actors and hazards. We show that learned affordance maps can be used to augment traditional approaches for both exploration and navigation, providing significant improvements in performance. |
| Tasks | Autonomous Vehicles |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BJgMFxrYPB |
| https://openreview.net/pdf?id=BJgMFxrYPB | |
| PWC | https://paperswithcode.com/paper/learning-to-move-with-affordance-maps |
| Repo | |
| Framework | |
AdvCodec: Towards A Unified Framework for Adversarial Text Generation
| Title | AdvCodec: Towards A Unified Framework for Adversarial Text Generation |
| Authors | Anonymous |
| Abstract | Machine learning (ML) especially deep neural networks (DNNs) have been widely applied to real-world applications. However, recent studies show that DNNs are vulnerable to carefully crafted \emph{adversarial examples} which only deviate from the original data by a small magnitude of perturbation. While there has been great interest on generating imperceptible adversarial examples in continuous data domain (e.g. image and audio) to explore the model vulnerabilities, generating \emph{adversarial text} in the discrete domain is still challenging. The main contribution of this paper is to propose a general targeted attack framework \advcodec for adversarial text generation which addresses the challenge of discrete input space and be easily adapted to general natural language processing (NLP) tasks. In particular, we propose a tree based autoencoder to encode discrete text data into continuous vector space, upon which we optimize the adversarial perturbation. With the tree based decoder, it is possible to ensure the grammar correctness of the generated text; and the tree based encoder enables flexibility of making manipulations on different levels of text, such as sentence (\advcodecsent) and word (\advcodecword) levels. We consider multiple attacking scenarios, including appending an adversarial sentence or adding unnoticeable words to a given paragraph, to achieve arbitrary \emph{targeted attack}. To demonstrate the effectiveness of the proposed method, we consider two most representative NLP tasks: sentiment analysis and question answering (QA). Extensive experimental results show that \advcodec has successfully attacked both tasks. In particular, our attack causes a BERT-based sentiment classifier accuracy to drop from $0.703$ to $0.006$, and a BERT-based QA model’s F1 score to drop from $88.62$ to $33.21$ (with best targeted attack F1 score as $46.54$). Furthermore, we show that the white-box generated adversarial texts can transfer across other black-box models, shedding light on an effective way to examine the robustness of existing NLP models. |
| Tasks | Adversarial Text, Question Answering, Sentiment Analysis, Text Generation |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rkeeoeHYvr |
| https://openreview.net/pdf?id=rkeeoeHYvr | |
| PWC | https://paperswithcode.com/paper/advcodec-towards-a-unified-framework-for |
| Repo | |
| Framework | |
Multi-task Network Embedding with Adaptive Loss Weighting
| Title | Multi-task Network Embedding with Adaptive Loss Weighting |
| Authors | Anonymous |
| Abstract | Network embedding is to learn low-dimensional representations of nodes which mostly preserve the network topological structure. In real-world networks, however, nodes are often associated with a rich set of attributes and labels which are potentially valuable in seeking more effective vector representations. To properly utilize this information, we propose a Joint Autoencoders framework for Multi-task network Embedding (JAME), which aims to encode a shared representation of local network structure, node attributes, and available node labels. Jointly em-bedding via multi-task learning is strongly dependent on the relative weighting between each task’s loss function. Tuning these weights by hand is an expensive and difficult process, making multi-task learning prohibitive in practice. Therefore, we define an adaptive loss weighting layer capable of learning an optimal combination of loss weights during representation learning. Empirical evaluations on real-world datasets show effectiveness and efficiency of our JAME model compared to relevant baseline methods. |
| Tasks | Multi-Task Learning, Network Embedding, Representation Learning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rkgTKerYwr |
| https://openreview.net/pdf?id=rkgTKerYwr | |
| PWC | https://paperswithcode.com/paper/multi-task-network-embedding-with-adaptive |
| Repo | |
| Framework | |
OPTIMAL TRANSPORT, CYCLEGAN, AND PENALIZED LS FOR UNSUPERVISED LEARNING IN INVERSE PROBLEMS
| Title | OPTIMAL TRANSPORT, CYCLEGAN, AND PENALIZED LS FOR UNSUPERVISED LEARNING IN INVERSE PROBLEMS |
| Authors | Anonymous |
| Abstract | The penalized least squares (PLS) is a classic approach to inverse problems, where a regularization term is added to stabilize the solution. Optimal transport (OT) is another mathematical framework for computer vision tasks by providing means to transport one measure to another at minimal cost. Cycle-consistent generative adversarial network (cycleGAN) is a recent extension of GAN to learn target distributions with less mode collapsing behavior. Although similar in that no supervised training is required, the algorithms look different, so the mathematical relationship between these approaches is not clear. In this article, we provide an important advance to unveil the missing link. Specifically, we reveal that a cycleGAN architecture can be derived as a dual formulation of the optimal transport problem, if the PLS with a deep learning penalty is used as a transport cost between the two probability measures from measurements and unknown images. This suggests that cycleGAN can be considered as stochastic generalization of classical PLS approaches. Our derivation is so general that various types of cycleGAN architecture can be easily derived by merely changing the transport cost. As proofs of concept, this paper provides novel cycleGAN architecture for unsupervised learning in accelerated MRI and deconvolution microscopy problems, which confirm the efficacy and the flexibility of the theory. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HklZUpEtvr |
| https://openreview.net/pdf?id=HklZUpEtvr | |
| PWC | https://paperswithcode.com/paper/optimal-transport-cyclegan-and-penalized-ls-1 |
| Repo | |
| Framework | |
Exploring the Correlation between Likelihood of Flow-based Generative Models and Image Semantics
| Title | Exploring the Correlation between Likelihood of Flow-based Generative Models and Image Semantics |
| Authors | Anonymous |
| Abstract | Among deep generative models, flow-based models, simply referred as \emph{flow}s in this paper, differ from other models in that they provide tractable likelihood. Besides being an evaluation metric of synthesized data, flows are supposed to be robust against out-of-distribution~(OoD) inputs since they do not discard any information of the inputs. However, it has been observed that flows trained on FashionMNIST assign higher likelihoods to OoD samples from MNIST. This counter-intuitive observation raises the concern about the robustness of flows’ likelihood. In this paper, we explore the correlation between flows’ likelihood and image semantics. We choose two typical flows as the target models: Glow, based on coupling transformations, and pixelCNN, based on autoregressive transformations. Our experiments reveal surprisingly weak correlation between flows’ likelihoods and image semantics: the predictive likelihoods of flows can be heavily affected by trivial transformations that keep the image semantics unchanged, which we call semantic-invariant transformations~(SITs). We explore three SITs~(all small pixel-level modifications): image pixel translation, random noise perturbation, latent factors zeroing~(limited to flows using multi-scale architecture, e.g. Glow). These findings, though counter-intuitive, resonate with the fact that the predictive likelihood of a flow is the joint probability of all the image pixels. So flows’ likelihoods, modeling on pixel-level intensities, is not able to indicate the existence likelihood of the high-level image semantics. We call for attention that it may be \emph{abuse} if we use the predictive likelihoods of flows for OoD samples detection. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rkgIllBtwB |
| https://openreview.net/pdf?id=rkgIllBtwB | |
| PWC | https://paperswithcode.com/paper/exploring-the-correlation-between-likelihood |
| Repo | |
| Framework | |
On the Weaknesses of Reinforcement Learning for Neural Machine Translation
| Title | On the Weaknesses of Reinforcement Learning for Neural Machine Translation |
| Authors | Anonymous |
| Abstract | Reinforcement learning (RL) is frequently used to increase performance in text generation tasks, including machine translation (MT), notably through the use of Minimum Risk Training (MRT) and Generative Adversarial Networks (GAN). However, little is known about what and how these methods learn in the context of MT. We prove that one of the most common RL methods for MT does not optimize the expected reward, as well as show that other methods take an infeasibly long time to converge. In fact, our results suggest that RL practices in MT are likely to improve performance only where the pre-trained parameters are already close to yielding the correct translation. Our findings further suggest that observed gains may be due to effects unrelated to the training signal, concretely, changes in the shape of the distribution curve. |
| Tasks | Machine Translation, Text Generation |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=H1eCw3EKvH |
| https://openreview.net/pdf?id=H1eCw3EKvH | |
| PWC | https://paperswithcode.com/paper/on-the-weaknesses-of-reinforcement-learning-1 |
| Repo | |
| Framework | |
D3PG: Deep Differentiable Deterministic Policy Gradients
| Title | D3PG: Deep Differentiable Deterministic Policy Gradients |
| Authors | Anonymous |
| Abstract | Over the last decade, two competing control strategies have emerged for solving complex control tasks with high efficacy. Model-based control algorithms, such as model-predictive control (MPC) and trajectory optimization, peer into the gradients of underlying system dynamics in order to solve control tasks with high sample efficiency. However, like all gradient-based numerical optimization methods,model-based control methods are sensitive to intializations and are prone to becoming trapped in local minima. Deep reinforcement learning (DRL), on the other hand, can somewhat alleviate these issues by exploring the solution space through sampling — at the expense of computational cost. In this paper, we present a hybrid method that combines the best aspects of gradient-based methods and DRL. We base our algorithm on the deep deterministic policy gradients (DDPG) algorithm and propose a simple modification that uses true gradients from a differentiable physical simulator to increase the convergence rate of both the actor and the critic. We demonstrate our algorithm on seven 2D robot control tasks, with the most complex one being a differentiable half cheetah with hard contact constraints. Empirical results show that our method boosts the performance of DDPGwithout sacrificing its robustness to local minima. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rkxZCJrtwS |
| https://openreview.net/pdf?id=rkxZCJrtwS | |
| PWC | https://paperswithcode.com/paper/d3pg-deep-differentiable-deterministic-policy |
| Repo | |
| Framework | |
Mixture Distributions for Scalable Bayesian Inference
| Title | Mixture Distributions for Scalable Bayesian Inference |
| Authors | Anonymous |
| Abstract | Bayesian Neural Networks (BNNs) provides a mathematically grounded framework to quantify uncertainty. However BNNs are computationally inefficient, thus are generally not employed on complicated machine learning tasks. Deep Ensembles were introduced as a Bootstrap inspired frequentist approach to the community, as an alternative to BNN’s. Ensembles of deterministic and stochastic networks are a good uncertainty estimator in various applications (Although, they are criticized for not being Bayesian). We show Ensembles of deterministic and stochastic Neural Networks can indeed be cast as an approximate Bayesian inference. Deep Ensembles have another weakness of having high space complexity, we provide an alternative to it by modifying the original Bayes by Backprop (BBB) algorithm to learn more general concrete mixture distributions over weights. We show our methods and its variants can give better uncertainty estimates at a significantly lower parametric overhead than Deep Ensembles. We validate our hypothesis through experiments like non-linear regression, predictive uncertainty estimation, detecting adversarial images and exploration-exploitation trade-off in reinforcement learning. |
| Tasks | Bayesian Inference |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=S1x6TlBtwB |
| https://openreview.net/pdf?id=S1x6TlBtwB | |
| PWC | https://paperswithcode.com/paper/mixture-distributions-for-scalable-bayesian |
| Repo | |
| Framework | |
On the geometry and learning low-dimensional embeddings for directed graphs
| Title | On the geometry and learning low-dimensional embeddings for directed graphs |
| Authors | Anonymous |
| Abstract | We propose a novel node embedding of directed graphs to statistical manifolds, which is based on a global minimization of pairwise relative entropy and graph geodesics in a non-linear way. Each node is encoded with a probability density function over a measurable space. Furthermore, we analyze the connection of the geometrical properties of such embedding and their efficient learning procedure. Extensive experiments show that our proposed embedding is better preserving the global geodesic information of graphs, as well as outperforming existing embedding models on directed graphs in a variety of evaluation metrics, in an unsupervised setting. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SkxQp1StDH |
| https://openreview.net/pdf?id=SkxQp1StDH | |
| PWC | https://paperswithcode.com/paper/on-the-geometry-and-learning-low-dimensional |
| Repo | |
| Framework | |
Is my Deep Learning Model Learning more than I want it to?
| Title | Is my Deep Learning Model Learning more than I want it to? |
| Authors | Anonymous |
| Abstract | Existing deep learning approaches for learning visual features tend to extract more information than what is required for the task at hand. From a privacy preservation perspective, the input visual information is not protected from the model; enabling the model to become more intelligent than it is trained to be. Existing approaches for suppressing additional task learning assume the presence of ground truth labels for the tasks to be suppressed during training time. In this research, we propose a three-fold novel contribution: (i) a novel metric to measure the trust score of a trained deep learning model, (ii) a model-agnostic solution framework for trust score improvement by suppressing all the unwanted tasks, and (iii) a simulated benchmark dataset, PreserveTask, having five different fundamental image classification tasks to study the generalization nature of models. In the first set of experiments, we measure and improve the trust scores of five popular deep learning models: VGG16, VGG19, Inception-v1, MobileNet, and DenseNet and demonstrate that Inception-v1 is having the lowest trust score. Additionally, we show results of our framework on color-MNIST dataset and practical applications of face attribute preservation in Diversity in Faces (DiF) and IMDB-Wiki dataset. |
| Tasks | Image Classification |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=B1lf4yBYPr |
| https://openreview.net/pdf?id=B1lf4yBYPr | |
| PWC | https://paperswithcode.com/paper/is-my-deep-learning-model-learning-more-than |
| Repo | |
| Framework | |
Stein Self-Repulsive Dynamics: Benefits from Past Samples
| Title | Stein Self-Repulsive Dynamics: Benefits from Past Samples |
| Authors | Anonymous |
| Abstract | We propose a new Stein self-repulsive dynamics for obtaining diversified samples from intractable un-normalized distributions. Our idea is to introduce Stein variational gradient as a repulsive force to push the samples of Langevin dynamics away from the past trajectories. This simple idea allows us to significantly decrease the auto-correlation in Langevin dynamics and hence increase the effective sample size. Importantly, as we establish in our theoretical analysis, the asymptotic stationary distribution remains correct even with the addition of the repulsive force, thanks to the special properties of the Stein variational gradient. We perform extensive empirical studies of our new algorithm, showing that our method yields much higher sample efficiency and better uncertainty estimation than vanilla Langevin dynamics. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=H1eWGREFvB |
| https://openreview.net/pdf?id=H1eWGREFvB | |
| PWC | https://paperswithcode.com/paper/stein-self-repulsive-dynamics-benefits-from |
| Repo | |
| Framework | |
GENESIS: Generative Scene Inference and Sampling with Object-Centric Latent Representations
| Title | GENESIS: Generative Scene Inference and Sampling with Object-Centric Latent Representations |
| Authors | Anonymous |
| Abstract | Generative latent-variable models are emerging as promising tools in robotics and reinforcement learning. Yet, even though tasks in these domains typically involve distinct objects, most state-of-the-art generative models do not explicitly capture the compositional nature of visual scenes. Two recent exceptions, MONet and IODINE, decompose scenes into objects in an unsupervised fashion. Their underlying generative processes, however, do not account for component interactions. Hence, neither of them allows for principled sampling of novel scenes. Here we present GENESIS, the first object-centric generative model of 3D visual scenes capable of both decomposing and generating scenes by capturing relationships between scene components. GENESIS parameterises a spatial GMM over images which is decoded from a set of object-centric latent variables that are either inferred sequentially in an amortised fashion or sampled from an autoregressive prior. We train GENESIS on several publicly available datasets and evaluate its performance on scene generation, decomposition, and semi-supervised learning. |
| Tasks | Latent Variable Models, Scene Generation |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BkxfaTVFwH |
| https://openreview.net/pdf?id=BkxfaTVFwH | |
| PWC | https://paperswithcode.com/paper/genesis-generative-scene-inference-and-1 |
| Repo | |
| Framework | |
