Uniformly Stable Algorithms for Adversarial Training and Beyond

arXiv:2405.01817v1 Announce Type: new
Abstract: In adversarial machine learning, neural networks suffer from a significant issue known as robust overfitting, where the robust test accuracy decreases over epochs (Rice et al., 2020). Recent research conducted by Xing et al.,2021; Xiao et al., 2022 has focused on studying the uniform stability of adversarial training. Their investigations revealed that SGD-based adversarial training fails to exhibit uniform stability, and the derived stability bounds align with the observed phenomenon of robust overfitting in experiments. …

abstract accuracy adversarial adversarial machine learning adversarial training algorithms arxiv beyond cs.lg issue machine machine learning networks neural networks overfitting research robust stability studying test training type uniform