Boosting Adversarial Training with Hypersphere Embedding

TL;DR

This paper introduces hypersphere embedding into adversarial training to improve model robustness against attacks, demonstrating consistent benefits across multiple frameworks and datasets with minimal additional computation.

Contribution

The paper proposes a novel integration of hypersphere embedding into adversarial training, enhancing robustness and compatibility with existing frameworks.

Findings

HE improves robustness across various AT methods

HE integration requires little extra computation

Enhanced robustness verified on CIFAR-10 and ImageNet

Abstract

Adversarial training (AT) is one of the most effective defenses against adversarial attacks for deep learning models. In this work, we advocate incorporating the hypersphere embedding (HE) mechanism into the AT procedure by regularizing the features onto compact manifolds, which constitutes a lightweight yet effective module to blend in the strength of representation learning. Our extensive analyses reveal that AT and HE are well coupled to benefit the robustness of the adversarially trained models from several aspects. We validate the effectiveness and adaptability of HE by embedding it into the popular AT frameworks including PGD-AT, ALP, and TRADES, as well as the FreeAT and FastAT strategies. In the experiments, we evaluate our methods under a wide range of adversarial attacks on the CIFAR-10 and ImageNet datasets, which verifies that integrating HE can consistently enhance the model robustness for each AT framework with little extra computation.