PixelDefend: Leveraging Generative Models to Understand and Defend against Adversarial Examples
Adversarial perturbations of normal images are usually imperceptible to humans, but they can seriously confuse state-of-the-art machine learning models. What makes them so special in the eyes of image classifiers? In this paper, we show empirically that adversarial examples mainly lie in the low probability regions of the training distribution, regardless of attack types and targeted models. Using statistical hypothesis testing, we find that modern neural density models are surprisingly good at detecting imperceptible image perturbations. Based on this discovery, we devised PixelDefend, a new approach that purifies a maliciously perturbed image by moving it back towards the distribution seen in the training data. The purified image is then run through an unmodified classifier, making our method agnostic to both the classifier and the attacking method. As a result, PixelDefend can be used to protect already deployed models and be combined with other model-specific defenses. Experiments show that our method greatly improves resilience across a wide variety of state-of-the-art attacking methods, increasing accuracy on the strongest attack from 63% to 84% for Fashion MNIST and from 32% to 70% for CIFAR-10.
Code (1)
Tasks
Two-sample testingSimilar Papers 제목 키워드 기반
Hilbert-Based Generative Defense for Adversarial Examples
Adversarial perturbations of clean images are usually imperceptible for human eyes, but can confidently fool deep neural networks (DNNs) to make incorrect predictions. Such vulnerability of DNNs raises serious security c…
AID-Purifier: A Light Auxiliary Network for Boosting Adversarial Defense
We propose an AID-purifier that can boost the robustness of adversarially-trained networks by purifying their inputs. AID-purifier is an auxiliary network that works as an add-on to an already trained main classifier. To…
Adversarial DefenseA Zero Trust Framework for Realization and Defense Against Generative AI Attacks in Power Grid
Understanding the potential of generative AI (GenAI)-based attacks on the power grid is a fundamental challenge that must be addressed in order to protect the power grid by realizing and validating risk in new attack vec…
Ensemble LearningDefending Against Unforeseen Failure Modes with Latent Adversarial Training
Despite extensive diagnostics and debugging by developers, AI systems sometimes exhibit harmful unintended behaviors. Finding and fixing these is challenging because the attack surface is so large -- it is not tractable …
image-classificationImage ClassificationRed Teamingtext-classification+2Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models
In recent years, deep neural network approaches have been widely adopted for machine learning tasks, including classification. However, they were shown to be vulnerable to adversarial perturbations: carefully crafted sma…
Adversarial DefenseAdversarial Defense against FGSM AttackGeneral Classification