arXiv subjects

Gaoyuan Zhang

Publications and source records attributed to Gaoyuan Zhang.

23 records · Page 2Linked to original sources

arXiv2020

Practical Detection of Trojan Neural Networks: Data-Limited and Data-Free Cases

When the training data are maliciously tampered, the predictions of the acquired deep neural network (DNN) can be manipulated by an adversary known as the Trojan attack (or poisoning backdoor attack). The lack of robustness of DNNs against Trojan attacks could significantly harm real-life machine learning (ML) systems in downstream applications, therefore posing widespread concern to their trustworthiness. In this paper, we study the problem of the Trojan network (TrojanNet) detection in the data-scarce regime, where only the weights of a trained DNN are accessed by the detector. We first propose a data-limited TrojanNet detector (TND), when only a few data samples are available for TrojanNet detection. We show that an effective data-limited TND can be established by exploring connections between Trojan attack and prediction-evasion adversarial attacks including per-sample attack as well as all-sample universal attack. In addition, we propose a data-free TND, which can detect a TrojanNet without accessing any data samples. We show that such a TND can be built by leveraging the internal response of hidden neurons, which exhibits the Trojan behavior even at random noise inputs. The effectiveness of our proposals is evaluated by extensive experiments under different model architectures and datasets including CIFAR-10, GTSRB, and ImageNet.

cs.LG↗

arXiv2020

Adversarial T-shirt! Evading Person Detectors in A Physical World

It is known that deep neural networks (DNNs) are vulnerable to adversarial attacks. The so-called physical adversarial examples deceive DNN-based decisionmakers by attaching adversarial patches to real objects. However, most of the existing works on physical adversarial attacks focus on static objects such as glass frames, stop signs and images attached to cardboard. In this work, we proposed adversarial T-shirts, a robust physical adversarial example for evading person detectors even if it could undergo non-rigid deformation due to a moving person's pose changes. To the best of our knowledge, this is the first work that models the effect of deformation for designing physical adversarial examples with respect to-rigid objects such as T-shirts. We show that the proposed method achieves74% and 57% attack success rates in the digital and physical worlds respectively against YOLOv2. In contrast, the state-of-the-art physical attack method to fool a person detector only achieves 18% attack success rate. Furthermore, by leveraging min-max optimization, we extend our method to the ensemble attack setting against two object detectors YOLO-v2 and Faster R-CNN simultaneously.

cs.CV↗

arXiv2020

A Primer on Zeroth-Order Optimization in Signal Processing and Machine Learning

Zeroth-order (ZO) optimization is a subset of gradient-free optimization that emerges in many signal processing and machine learning applications. It is used for solving optimization problems similarly to gradient-based methods. However, it does not require the gradient, using only function evaluations. Specifically, ZO optimization iteratively performs three major steps: gradient estimation, descent direction computation, and solution update. In this paper, we provide a comprehensive review of ZO optimization, with an emphasis on showing the underlying intuition, optimization principles and recent advances in convergence analysis. Moreover, we demonstrate promising applications of ZO optimization, such as evaluating robustness and generating explanations from black-box deep learning models, and efficient online sensor management.

cs.LG↗

arXiv2019

Reflecting After Learning for Understanding

Today, image classification is a common way for systems to process visual content. Although neural network approaches to classification have seen great progress in reducing error rates, it is not clear what this means for a cognitive system that needs to make sense of the multiple and competing predictions from its own classifiers. As a step to address this, we present a novel framework that uses meta-reasoning and meta-operations to unify predictions into abstractions, properties, or relationships. Using the framework on images from ImageNet, we demonstrate systems that unify 41% to 46% of predictions in general and unify 67% to 75% of predictions when the systems can explain their conceptual differences. We also demonstrate a system in "the wild" by feeding live video images through it and show it unifying 51% of predictions in general and 69% of predictions when their differences can be explained conceptually by the system. In a survey given to 24 participants, we found that 87% of the unified predictions describe their corresponding images.

cs.AI↗

arXiv2019

Interpreting Adversarial Examples by Activation Promotion and Suppression

It is widely known that convolutional neural networks (CNNs) are vulnerable to adversarial examples: images with imperceptible perturbations crafted to fool classifiers. However, interpretability of these perturbations is less explored in the literature. This work aims to better understand the roles of adversarial perturbations and provide visual explanations from pixel, image and network perspectives. We show that adversaries have a promotion-suppression effect (PSE) on neurons' activations and can be primarily categorized into three types: i) suppression-dominated perturbations that mainly reduce the classification score of the true label, ii) promotion-dominated perturbations that focus on boosting the confidence of the target label, and iii) balanced perturbations that play a dual role in suppression and promotion. We also provide image-level interpretability of adversarial examples. This links PSE of pixel-level perturbations to class-specific discriminative image regions localized by class activation mapping (Zhou et al. 2016). Further, we examine the adversarial effect through network dissection (Bau et al. 2017), which offers concept-level interpretability of hidden units. We show that there exists a tight connection between the units' sensitivity to adversarial attacks and their interpretability on semantic concepts. Lastly, we provide some new insights from our interpretation to improve the adversarial robustness of networks.

cs.CV↗