April 1, 2020

2953 words 14 mins read

Paper Group NANR 16

Paper Group NANR 16

Compressive Recovery Defense: A Defense Framework for $\ell_0, \ell_2$ and $\ell_\infty$ norm attacks.. Exploring Cellular Protein Localization Through Semantic Image Synthesis. Latent Variables on Spheres for Sampling and Inference. Sequence-level Intrinsic Exploration Model for Partially Observable Domains. Perceptual Regularization: Visualizing …

Compressive Recovery Defense: A Defense Framework for $\ell_0, \ell_2$ and $\ell_\infty$ norm attacks.

Title Compressive Recovery Defense: A Defense Framework for $\ell_0, \ell_2$ and $\ell_\infty$ norm attacks.
Authors Anonymous
Abstract We provide recovery guarantees for compressible signals that have been corrupted with noise and extend the framework introduced in \cite{bafna2018thwarting} to defend neural networks against $\ell_0$, $\ell_2$, and $\ell_{\infty}$-norm attacks. In the case of $\ell_0$-norm noise, we provide recovery guarantees for Iterative Hard Thresholding (IHT) and Basis Pursuit (BP). For $\ell_2$-norm bounded noise, we provide recovery guarantees for BP, and for the case of $\ell_\infty$-norm bounded noise, we provide recovery guarantees for Dantzig Selector (DS). These guarantees theoretically bolster the defense framework introduced in \cite{bafna2018thwarting} for defending neural networks against adversarial inputs. Finally, we experimentally demonstrate the effectiveness of this defense framework against an array of $\ell_0$, $\ell_2$ and $\ell_\infty$-norm attacks.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=B1x9ITVYDr
PDF https://openreview.net/pdf?id=B1x9ITVYDr
PWC https://paperswithcode.com/paper/compressive-recovery-defense-a-defense
Repo
Framework

Exploring Cellular Protein Localization Through Semantic Image Synthesis

Title Exploring Cellular Protein Localization Through Semantic Image Synthesis
Authors Anonymous
Abstract Cell-cell interactions have an integral role in tumorigenesis as they are critical in governing immune responses. As such, investigating specific cell-cell interactions has the potential to not only expand upon the understanding of tumorigenesis, but also guide clinical management of patient responses to cancer immunotherapies. A recent imaging technique for exploring cell-cell interactions, multiplexed ion beam imaging by time-of-flight (MIBI-TOF), allows for cells to be quantified in 36 different protein markers at sub-cellular resolutions in situ as high resolution multiplexed images. To explore the MIBI images, we propose a GAN for multiplexed data with protein specific attention. By conditioning image generation on cell types, sizes, and neighborhoods through semantic segmentation maps, we are able to observe how these factors affect cell-cell interactions simultaneously in different protein channels. Furthermore, we design a set of metrics and offer the first insights towards cell spatial orientations, cell protein expressions, and cell neighborhoods. Our model, cell-cell interaction GAN (CCIGAN), outperforms or matches existing image synthesis methods on all conventional measures and significantly outperforms on biologically motivated metrics. To our knowledge, we are the first to systematically model multiple cellular protein behaviors and interactions under simulated conditions through image synthesis.
Tasks Image Generation, Semantic Segmentation
Published 2020-01-01
URL https://openreview.net/forum?id=Byg9AR4YDB
PDF https://openreview.net/pdf?id=Byg9AR4YDB
PWC https://paperswithcode.com/paper/exploring-cellular-protein-localization
Repo
Framework

Latent Variables on Spheres for Sampling and Inference

Title Latent Variables on Spheres for Sampling and Inference
Authors Anonymous
Abstract Variational inference is a fundamental problem in Variational AutoEncoder (VAE). The optimization with lower bound of marginal log-likelihood results in the distribution of latent variables approximate to a given prior probability, which is the dilemma of employing VAE to solve real-world problems. By virtue of high-dimensional geometry, we propose a very simple algorithm completely different from existing ones to alleviate the variational inference in VAE. We analyze the unique characteristics of random variables on spheres in high dimensions and prove that Wasserstein distance between two arbitrary data sets randomly drawn from a sphere are nearly identical when the dimension is sufficiently large. Based on our theory, a novel algorithm for distribution-robust sampling is devised. Moreover, we reform the latent space of VAE by constraining latent variables on the sphere, thus freeing VAE from the approximate optimization of posterior probability via variational inference. The new algorithm is named Spherical AutoEncoder (SAE). Extensive experiments by sampling and inference tasks validate our theoretical analysis and the superiority of SAE.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=rJx2slSKDS
PDF https://openreview.net/pdf?id=rJx2slSKDS
PWC https://paperswithcode.com/paper/latent-variables-on-spheres-for-sampling-and
Repo
Framework

Sequence-level Intrinsic Exploration Model for Partially Observable Domains

Title Sequence-level Intrinsic Exploration Model for Partially Observable Domains
Authors Anonymous
Abstract Training reinforcement learning policies in partially observable domains with sparse reward signal is an important and open problem for the research community. In this paper, we introduce a new sequence-level intrinsic novelty model to tackle the challenge of training reinforcement learning policies in sparse rewarded partially observable domains. First, we propose a new reasoning paradigm to infer the novelty for the partially observable states, which is built upon forward dynamics prediction. Different from conventional approaches that perform self-prediction or one-step forward prediction, our proposed approach engages open-loop multi-step prediction, which enables the difficulty of novelty prediction to flexibly scale and thus results in high-quality novelty scores. Second, we propose a novel dual-LSTM architecture to facilitate the sequence-level reasoning over the partially observable state space. Our proposed architecture efficiently synthesizes information from an observation sequence and an action sequence to derive meaningful latent representations for inferring the novelty for states. To evaluate the efficiency of our proposed approach, we conduct extensive experiments on several challenging 3D navigation tasks from ViZDoom and DeepMind Lab. We also present results on two hard-exploration domains from Atari 2600 series in Appendix to demonstrate our proposed approach could generalize beyond partially observable navigation tasks. Overall, the experiment results reveal that our proposed intrinsic novelty model could outperform several state-of-the-art curiosity baselines with considerable significance in the testified domains.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=H1eCR34FPB
PDF https://openreview.net/pdf?id=H1eCR34FPB
PWC https://paperswithcode.com/paper/sequence-level-intrinsic-exploration-model
Repo
Framework

Perceptual Regularization: Visualizing and Learning Generalizable Representations

Title Perceptual Regularization: Visualizing and Learning Generalizable Representations
Authors Anonymous
Abstract A deployable machine learning model relies on a good representation. Two desirable criteria of a good representation are to be understandable, and to generalize to new tasks. We propose a technique termed perceptual regularization that enables both visualization of the latent representation and control over the generality of the learned representation. In particular our method provides a direct visualization of the effect that adversarial attacks have on the internal representation of a deep network. By visualizing the learned representation, we are also able to understand the attention of a model, obtaining visual evidence that supervised networks learn task-specific representations. We show models trained with perceptual regularization learn transferrable features, achieving significantly higher accuracy in unseen tasks compared to standard supervised learning and multi-task methods.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=S1e1EAEFPB
PDF https://openreview.net/pdf?id=S1e1EAEFPB
PWC https://paperswithcode.com/paper/perceptual-regularization-visualizing-and
Repo
Framework

3D Human Pose Estimation via Explicit Compositional Depth Maps

Title 3D Human Pose Estimation via Explicit Compositional Depth Maps
Authors Haiping Wu, Bin Xiao
Abstract n this work, we tackle the problem of estimating 3D human pose in camera space from a monocular image. First, we propose to use densely-generated limb depth maps to ease the learning of body joints depth, which are well aligned with image cues. Then, we design a lifting module from 2D pixel coordinates to 3D camera coordinates which explicitly takes the depth values as inputs, and is aligned with camera perspective projection model. We show our method achieves superior performance on large-scale 3D pose datasets Human3.6M and MPI-INF-3DHP, and sets the new state-of-the-art.
Tasks 3D Human Pose Estimation, Pose Estimation
Published 2020-02-08
URL https://www.researchgate.net/publication/339375538_3D_Human_Pose_Estimation_via_Explicit_Compositional_Depth_Maps
PDF https://www.researchgate.net/publication/339375538_3D_Human_Pose_Estimation_via_Explicit_Compositional_Depth_Maps
PWC https://paperswithcode.com/paper/3d-human-pose-estimation-via-explicit
Repo
Framework

Rethinking Data Augmentation: Self-Supervision and Self-Distillation

Title Rethinking Data Augmentation: Self-Supervision and Self-Distillation
Authors Anonymous
Abstract Data augmentation techniques, e.g., flipping or cropping, which systematically enlarge the training dataset by explicitly generating more training samples, are effective in improving the generalization performance of deep neural networks. In the supervised setting, a common practice for data augmentation is to assign the same label to all augmented samples of the same source. However, if the augmentation results in large distributional discrepancy among them (e.g., rotations), forcing their label invariance may be too difficult to solve and often hurts the performance. To tackle this challenge, we suggest a simple yet effective idea of learning the joint distribution of the original and self-supervised labels of augmented samples. The joint learning framework is easier to train, and enables an aggregated inference combining the predictions from different augmented samples for improving the performance. Further, to speed up the aggregation process, we also propose a knowledge transfer technique, self-distillation, which transfers the knowledge of augmentation into the model itself. We demonstrate the effectiveness of our data augmentation framework on various fully-supervised settings including the few-shot and imbalanced classification scenarios.
Tasks Data Augmentation, Transfer Learning
Published 2020-01-01
URL https://openreview.net/forum?id=SkliR1SKDS
PDF https://openreview.net/pdf?id=SkliR1SKDS
PWC https://paperswithcode.com/paper/rethinking-data-augmentation-self-supervision
Repo
Framework

LAMAL: LAnguage Modeling Is All You Need for Lifelong Language Learning

Title LAMAL: LAnguage Modeling Is All You Need for Lifelong Language Learning
Authors Anonymous
Abstract Most research on lifelong learning (LLL) applies to images or games, but not language. We present LAMAL, a simple yet effective method for LLL based on language modeling. LAMAL replays pseudo-samples of previous tasks while requiring no extra memory or model capacity. Specifically, LAMAL is a language model that simultaneously learns to solve the task and generate training samples. When the model is trained for a new task, it generates pseudo-samples of previous tasks for training alongside data for the new task. The results show that LAMAL prevents catastrophic forgetting without any sign of intransigence and can perform up to five very different language tasks sequentially with only one model. Overall, LAMAL outperforms previous methods by a considerable margin and is only 2–3% worse than multitasking, which is usually considered the LLL upper bound. The source code is available at https://github.com/xxx.
Tasks Language Modelling
Published 2020-01-01
URL https://openreview.net/forum?id=Skgxcn4YDS
PDF https://openreview.net/pdf?id=Skgxcn4YDS
PWC https://paperswithcode.com/paper/lamal-language-modeling-is-all-you-need-for-1
Repo
Framework

Variational Information Bottleneck for Unsupervised Clustering: Deep Gaussian Mixture Embedding

Title Variational Information Bottleneck for Unsupervised Clustering: Deep Gaussian Mixture Embedding
Authors Anonymous
Abstract In this paper, we develop an unsupervised generative clustering framework that combines variational information bottleneck and the Gaussian Mixture Model. Specifically, in our approach we use the variational information bottleneck method and model the latent space as a mixture of Gaussians. We derive a bound on the cost function of our model that generalizes the evidence lower bound (ELBO); and provide a variational inference type algorithm that allows to compute it. In the algorithm, the coders’ mappings are parametrized using neural networks and the bound is approximated by Markov sampling and optimized with stochastic gradient descent. Numerical results on real datasets are provided to support the efficiency of our method.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=HyxQ3gSKvr
PDF https://openreview.net/pdf?id=HyxQ3gSKvr
PWC https://paperswithcode.com/paper/variational-information-bottleneck-for-1
Repo
Framework

Unsupervised Domain Adaptation through Self-Supervision

Title Unsupervised Domain Adaptation through Self-Supervision
Authors Anonymous
Abstract This paper addresses unsupervised domain adaptation, the setting where labeled training data is available on a source domain, but the goal is to have good performance on a target domain with only unlabeled data. Like much of previous work, we seek to align the learned representations of the source and target domains while preserving discriminability. The way we accomplish alignment is by learning to perform auxiliary self-supervised task(s) on both domains simultaneously. Each self-supervised task brings the two domains closer together along the direction relevant to that task. Training this jointly with the main task classifier on the source domain is shown to successfully generalize to the unlabeled target domain. The presented objective is straightforward to implement and easy to optimize. We achieve state-of-the-art results on four out of seven standard benchmarks, and competitive results on segmentation adaptation. We also demonstrate that our method composes well with another popular pixel-level adaptation method.
Tasks Domain Adaptation, Unsupervised Domain Adaptation
Published 2020-01-01
URL https://openreview.net/forum?id=S1lF8xHYwS
PDF https://openreview.net/pdf?id=S1lF8xHYwS
PWC https://paperswithcode.com/paper/unsupervised-domain-adaptation-through-self
Repo
Framework

Global Momentum Compression for Sparse Communication in Distributed SGD

Title Global Momentum Compression for Sparse Communication in Distributed SGD
Authors Anonymous
Abstract With the rapid growth of data, distributed stochastic gradient descent~(DSGD) has been widely used for solving large-scale machine learning problems. Due to the latency and limited bandwidth of network, communication has become the bottleneck of DSGD when we need to train large scale models, like deep neural networks. Communication compression with sparsified gradient, abbreviated as \emph{sparse communication}, has been widely used for reducing communication cost in DSGD. Recently, there has appeared one method, called deep gradient compression~(DGC), to combine memory gradient and momentum SGD for sparse communication. DGC has achieved promising performance in practice. However, the theory about the convergence of DGC is lack. In this paper, we propose a novel method, called \emph{\underline{g}}lobal \emph{\underline{m}}omentum \emph{\underline{c}}ompression~(GMC), for sparse communication in DSGD. GMC also combines memory gradient and momentum SGD. But different from DGC which adopts local momentum, GMC adopts global momentum. We theoretically prove the convergence rate of GMC for both convex and non-convex problems. To the best of our knowledge, this is the first work that proves the convergence of distributed momentum SGD~(DMSGD) with sparse communication and memory gradient. Empirical results show that, compared with the DMSGD counterpart without sparse communication, GMC can reduce the communication cost by approximately 100 fold without loss of generalization accuracy. GMC can also achieve comparable~(sometimes better) performance compared with DGC, with an extra theoretical guarantee.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=ryedjkSFwr
PDF https://openreview.net/pdf?id=ryedjkSFwr
PWC https://paperswithcode.com/paper/global-momentum-compression-for-sparse-1
Repo
Framework

Improving Model Compatibility of Generative Adversarial Networks by Boundary Calibration

Title Improving Model Compatibility of Generative Adversarial Networks by Boundary Calibration
Authors Anonymous
Abstract Generative Adversarial Networks (GANs) is a powerful family of models that learn an underlying distribution to generate synthetic data. Many existing studies of GANs focus on improving the realness of the generated image data for visual applications, and few of them concern about improving the quality of the generated data for training other classifiers—a task known as the model compatibility problem. As a consequence, existing GANs often prefer generating `easier’ synthetic data that are far from the boundaries of the classifiers, and refrain from generating near-boundary data, which are known to play an important roles in training the classifiers. To improve GAN in terms of model compatibility, we propose Boundary-Calibration GANs (BCGANs), which leverage the boundary information from a set of pre-trained classifiers using the original data. In particular, we introduce an auxiliary Boundary-Calibration loss (BC-loss) into the generator of GAN to match the statistics between the posterior distributions of original data and generated data with respect to the boundaries of the pre-trained classifiers. The BC-loss is provably unbiased and can be easily coupled with different GAN variants to improve their model compatibility. Experimental results demonstrate that BCGANs not only generate realistic images like original GANs but also achieves superior model compatibility than the original GANs. |
Tasks Calibration
Published 2020-01-01
URL https://openreview.net/forum?id=S1xJikHtDH
PDF https://openreview.net/pdf?id=S1xJikHtDH
PWC https://paperswithcode.com/paper/improving-model-compatibility-of-generative
Repo
Framework

Meta-RCNN: Meta Learning for Few-Shot Object Detection

Title Meta-RCNN: Meta Learning for Few-Shot Object Detection
Authors Anonymous
Abstract Despite significant advances in object detection in recent years, training effective detectors in a small data regime remains an open challenge. Labelling training data for object detection is extremely expensive, and there is a need to develop techniques that can generalize well from small amounts of labelled data. We investigate this problem of few-shot object detection, where a detector has access to only limited amounts of annotated data. Based on the recently evolving meta-learning principle, we propose a novel meta-learning framework for object detection named ``Meta-RCNN”, which learns the ability to perform few-shot detection via meta-learning. Specifically, Meta-RCNN learns an object detector in an episodic learning paradigm on the (meta) training data. This learning scheme helps acquire a prior which enables Meta-RCNN to do few-shot detection on novel tasks. Built on top of the Faster RCNN model, in Meta-RCNN, both the Region Proposal Network (RPN) and the object classification branch are meta-learned. The meta-trained RPN learns to provide class-specific proposals, while the object classifier learns to do few-shot classification. The novel loss objectives and learning strategy of Meta-RCNN can be trained in an end-to-end manner. We demonstrate the effectiveness of Meta-RCNN in addressing few-shot detection on Pascal VOC dataset and achieve promising results. |
Tasks Few-Shot Object Detection, Meta-Learning, Object Classification, Object Detection
Published 2020-01-01
URL https://openreview.net/forum?id=B1xmOgrFPS
PDF https://openreview.net/pdf?id=B1xmOgrFPS
PWC https://paperswithcode.com/paper/meta-rcnn-meta-learning-for-few-shot-object
Repo
Framework
Title Neural Operator Search
Authors Anonymous
Abstract Existing neural architecture search (NAS) methods explore a limited feature-transformation-only search space while ignoring other advanced feature operations such as feature self-calibration by attention and dynamic convolutions. This disables the NAS algorithms to discover more advanced network architectures. We address this limitation by additionally exploiting feature self-calibration operations, resulting in a heterogeneous search space. To solve the challenges of operation heterogeneity and significantly larger search space, we formulate a neural operator search (NOS) method. NOS presents a novel heterogeneous residual block for integrating the heterogeneous operations in a unified structure, and an attention guided search strategy for facilitating the search process over a vast space. Extensive experiments show that NOS can search novel cell architectures with highly competitive performance on the CIFAR and ImageNet benchmarks.
Tasks Calibration, Neural Architecture Search
Published 2020-01-01
URL https://openreview.net/forum?id=H1lxeRNYvB
PDF https://openreview.net/pdf?id=H1lxeRNYvB
PWC https://paperswithcode.com/paper/neural-operator-search
Repo
Framework

Model-based Saliency for the Detection of Adversarial Examples

Title Model-based Saliency for the Detection of Adversarial Examples
Authors Anonymous
Abstract Adversarial perturbations cause a shift in the salient features of an image, which may result in a misclassification. We demonstrate that gradient-based saliency approaches are unable to capture this shift, and develop a new defense which detects adversarial examples based on learnt saliency models instead. We study two approaches: a CNN trained to distinguish between natural and adversarial images using the saliency masks produced by our learnt saliency model, and a CNN trained on the salient pixels themselves as its input. On MNIST, CIFAR-10 and ASSIRA, our defenses are able to detect various adversarial attacks, including strong attacks such as C&W and DeepFool, contrary to gradient-based saliency and detectors which rely on the input image. The latter are unable to detect adversarial images when the L_2- and L_infinity- norms of the perturbations are too small. Lastly, we find that the salient pixel based detector improves on saliency map based detectors as it is more robust to white-box attacks.
Tasks
Published 2020-01-01
URL https://openreview.net/forum?id=HJe5_6VKwS
PDF https://openreview.net/pdf?id=HJe5_6VKwS
PWC https://paperswithcode.com/paper/model-based-saliency-for-the-detection-of
Repo
Framework
comments powered by Disqus