Semantics-Preserving Adversarial Training

Abstract

Adversarial training is a defense technique that improves adversarial robustness of a deep neural network (DNN) by including adversarial examples in the training data. In this paper, we identify an overlooked problem of adversarial training in that these adversarial examples often have different semantics than the original data, introducing unintended biases into the model. We hypothesize that such non-semantics-preserving (and resultingly ambiguous) adversarial data harm the robustness of the target models. To mitigate such unintended semantic changes of adversarial examples, we propose semantics-preserving adversarial training (SPAT) which encourages perturbation on the pixels that are shared among all classes when generating adversarial examples in the training stage. Experiment results show that SPAT improves adversarial robustness and achieves state-of-the-art results in CIFAR-10 and CIFAR-100.

Preprint. Under Review

References in corpus (10)

• Wide Residual Networks
• Theoretically Principled Trade-off between Robustness and Accuracy
• Provable defenses against adversarial examples via the convex outer adversarial polytope
• Certified Adversarial Robustness via Randomized Smoothing
• Adversarial Risk and the Dangers of Evaluating Against Weak Attacks
• Improving the Adversarial Robustness and Interpretability of Deep Neural Networks by Regularizing their Input Gradients
• Exploring the Landscape of Spatial Robustness
• Smooth Adversarial Training
• Interpreting Adversarially Trained Convolutional Neural Networks
• Metric Learning for Adversarial Robustness