Adversarial training is one of the most effective approaches to improve model robustness against adversarial examples. However, previous works mainly focus on the overall robustness of the model, and the in-depth analysis on the role of each class involved in adversarial training is still missing. In this paper, we propose to analyze the class-wise robustness in adversarial training. First, we provide a detailed diagnosis of adversarial training on six benchmark datasets, i.e., MNIST, CIFAR-10, CIFAR-100, SVHN, STL-10 and ImageNet. Surprisingly, we find that there are remarkable robustness discrepancies among classes, leading to unbalance/unfair class-wise robustness in the robust models. Furthermore, we keep investigating the relations between classes and find that the unbalanced class-wise robustness is pretty consistent among different attack and defense methods. Moreover, we observe that the stronger attack methods in adversarial learning achieve performance improvement mainly from a more successful attack on the vulnerable classes (i.e., classes with less robustness). Inspired by these interesting findings, we design a simple but effective attack method based on the traditional PGD attack, named Temperature-PGD attack, which proposes to enlarge the robustness disparity among classes with a temperature factor on the confidence distribution of each image. Experiments demonstrate our method can achieve a higher attack rate than the PGD attack. Furthermore, from the defense perspective, we also make some modifications in the training and inference phase to improve the robustness of the most vulnerable class, so as to mitigate the large difference in class-wise robustness. We believe our work can contribute to a more comprehensive understanding of adversarial training as well as rethinking the class-wise properties in robust models.
相關內容
Adversarial robustness is considered as a required property of deep neural networks. In this study, we discover that adversarially trained models might have significantly different characteristics in terms of margin and smoothness, even they show similar robustness. Inspired by the observation, we investigate the effect of different regularizers and discover the negative effect of the smoothness regularizer on maximizing the margin. Based on the analyses, we propose a new method called bridged adversarial training that mitigates the negative effect by bridging the gap between clean and adversarial examples. We provide theoretical and empirical evidence that the proposed method provides stable and better robustness, especially for large perturbations.
Despite achieving strong performance in semi-supervised node classification task, graph neural networks (GNNs) are vulnerable to adversarial attacks, similar to other deep learning models. Existing researches focus on developing either robust GNN models or attack detection methods against adversarial attacks on graphs. However, little research attention is paid to the potential and practice of immunization to adversarial attacks on graphs. In this paper, we propose and formulate the graph adversarial immunization problem, i.e., vaccinating an affordable fraction of node pairs, connected or unconnected, to improve the certifiable robustness of graph against any admissible adversarial attack. We further propose an effective algorithm, called AdvImmune, which optimizes with meta-gradient in a discrete way to circumvent the computationally expensive combinatorial optimization when solving the adversarial immunization problem. Experiments are conducted on two citation networks and one social network. Experimental results demonstrate that the proposed AdvImmune method remarkably improves the ratio of robust nodes by 12%, 42%, 65%, with an affordable immune budget of only 5% edges.
Recent neural-based relation extraction approaches, though achieving promising improvement on benchmark datasets, have reported their vulnerability towards adversarial attacks. Thus far, efforts mostly focused on generating adversarial samples or defending adversarial attacks, but little is known about the difference between normal and adversarial samples. In this work, we take the first step to leverage the salience-based method to analyze those adversarial samples. We observe that salience tokens have a direct correlation with adversarial perturbations. We further find the adversarial perturbations are either those tokens not existing in the training set or superficial cues associated with relation labels. To some extent, our approach unveils the characters against adversarial samples. We release an open-source testbed, "DiagnoseAdv".
Upon the discovery of adversarial attacks, robust models have become obligatory for deep learning-based systems. Adversarial training with first-order attacks has been one of the most effective defenses against adversarial perturbations to this day. The majority of the adversarial training approaches focus on iteratively perturbing each pixel with the gradient of the loss function with respect to the input image. However, the adversarial training with gradient-based attacks lacks diversity and does not generalize well to natural images and various attacks. This study presents a robust training algorithm where the adversarial perturbations are automatically synthesized from a random vector using a generator network. The classifier is trained with cross-entropy loss regularized with the optimal transport distance between the representations of the natural and synthesized adversarial samples. Unlike prevailing generative defenses, the proposed one-step attack generation framework synthesizes diverse perturbations without utilizing gradient of the classifier's loss. Experimental results show that the proposed approach attains comparable robustness with various gradient-based and generative robust training techniques on CIFAR10, CIFAR100, and SVHN datasets. In addition, compared to the baselines, the proposed robust training framework generalizes well to the natural samples. Code and trained models will be made publicly available.
In S&P '21, Jia et al. proposed a new concept/mechanism named proof-of-learning (PoL), which allows a prover to demonstrate ownership of a machine learning model by proving integrity of the training procedure. It guarantees that an adversary cannot construct a valid proof with less cost (in both computation and storage) than that made by the prover in generating the proof. A PoL proof includes a set of intermediate models recorded during training, together with the corresponding data points used to obtain each recorded model. Jia et al. claimed that an adversary merely knowing the final model and training dataset cannot efficiently find a set of intermediate models with correct data points. In this paper, however, we show that PoL is vulnerable to "adversarial examples"! Specifically, in a similar way as optimizing an adversarial example, we could make an arbitrarily-chosen data point "generate" a given model, hence efficiently generating intermediate models with correct data points. We demonstrate, both theoretically and empirically, that we are able to generate a valid proof with significantly less cost than generating a proof by the prover, thereby we successfully break PoL.
While existing work in robust deep learning has focused on small pixel-level $\ell_p$ norm-based perturbations, this may not account for perturbations encountered in several real world settings. In many such cases although test data might not be available, broad specifications about the types of perturbations (such as an unknown degree of rotation) may be known. We consider a setup where robustness is expected over an unseen test domain that is not i.i.d. but deviates from the training domain. While this deviation may not be exactly known, its broad characterization is specified a priori, in terms of attributes. We propose an adversarial training approach which learns to generate new samples so as to maximize exposure of the classifier to the attributes-space, without having access to the data from the test domain. Our adversarial training solves a min-max optimization problem, with the inner maximization generating adversarial perturbations, and the outer minimization finding model parameters by optimizing the loss on adversarial perturbations generated from the inner maximization. We demonstrate the applicability of our approach on three types of naturally occurring perturbations -- object-related shifts, geometric transformations, and common image corruptions. Our approach enables deep neural networks to be robust against a wide range of naturally occurring perturbations. We demonstrate the usefulness of the proposed approach by showing the robustness gains of deep neural networks trained using our adversarial training on MNIST, CIFAR-10, and a new variant of the CLEVR dataset.
Deep learning models on graphs have achieved remarkable performance in various graph analysis tasks, e.g., node classification, link prediction and graph clustering. However, they expose uncertainty and unreliability against the well-designed inputs, i.e., adversarial examples. Accordingly, various studies have emerged for both attack and defense addressed in different graph analysis tasks, leading to the arms race in graph adversarial learning. For instance, the attacker has poisoning and evasion attack, and the defense group correspondingly has preprocessing- and adversarial- based methods. Despite the booming works, there still lacks a unified problem definition and a comprehensive review. To bridge this gap, we investigate and summarize the existing works on graph adversarial learning tasks systemically. Specifically, we survey and unify the existing works w.r.t. attack and defense in graph analysis tasks, and give proper definitions and taxonomies at the same time. Besides, we emphasize the importance of related evaluation metrics, and investigate and summarize them comprehensively. Hopefully, our works can serve as a reference for the relevant researchers, thus providing assistance for their studies. More details of our works are available at //github.com/gitgiter/Graph-Adversarial-Learning.
Deep neural networks (DNN) have achieved unprecedented success in numerous machine learning tasks in various domains. However, the existence of adversarial examples has raised concerns about applying deep learning to safety-critical applications. As a result, we have witnessed increasing interests in studying attack and defense mechanisms for DNN models on different data types, such as images, graphs and text. Thus, it is necessary to provide a systematic and comprehensive overview of the main threats of attacks and the success of corresponding countermeasures. In this survey, we review the state of the art algorithms for generating adversarial examples and the countermeasures against adversarial examples, for the three popular data types, i.e., images, graphs and text.
Generating high-quality and interpretable adversarial examples in the text domain is a much more daunting task than it is in the image domain. This is due partly to the discrete nature of text, partly to the problem of ensuring that the adversarial examples are still probable and interpretable, and partly to the problem of maintaining label invariance under input perturbations. In order to address some of these challenges, we introduce sparse projected gradient descent (SPGD), a new approach to crafting interpretable adversarial examples for text. SPGD imposes a directional regularization constraint on input perturbations by projecting them onto the directions to nearby word embeddings with highest cosine similarities. This constraint ensures that perturbations move each word embedding in an interpretable direction (i.e., towards another nearby word embedding). Moreover, SPGD imposes a sparsity constraint on perturbations at the sentence level by ignoring word-embedding perturbations whose norms are below a certain threshold. This constraint ensures that our method changes only a few words per sequence, leading to higher quality adversarial examples. Our experiments with the IMDB movie review dataset show that the proposed SPGD method improves adversarial example interpretability and likelihood (evaluated by average per-word perplexity) compared to state-of-the-art methods, while suffering little to no loss in training performance.
Adversarial attacks to image classification systems present challenges to convolutional networks and opportunities for understanding them. This study suggests that adversarial perturbations on images lead to noise in the features constructed by these networks. Motivated by this observation, we develop new network architectures that increase adversarial robustness by performing feature denoising. Specifically, our networks contain blocks that denoise the features using non-local means or other filters; the entire networks are trained end-to-end. When combined with adversarial training, our feature denoising networks substantially improve the state-of-the-art in adversarial robustness in both white-box and black-box attack settings. On ImageNet, under 10-iteration PGD white-box attacks where prior art has 27.9% accuracy, our method achieves 55.7%; even under extreme 2000-iteration PGD white-box attacks, our method secures 42.6% accuracy. A network based on our method was ranked first in Competition on Adversarial Attacks and Defenses (CAAD) 2018 --- it achieved 50.6% classification accuracy on a secret, ImageNet-like test dataset against 48 unknown attackers, surpassing the runner-up approach by ~10%. Code and models will be made publicly available.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191