# Stability Analysis and Generalization Bounds of Adversarial Training

Source: [Stability Analysis and Generalization Bounds of Adversarial Training](<https://www.sota2.com/research/paper/68fa6dc81e209131cea3c30c>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-12-24T17:58:09.085Z

Authors: Jiancong Xiao, Yanbo Fan, Ruoyu Sun, Jue Wang, Zhi-Quan Luo

Published: 2022-10-03T14:21:46.000Z

## Abstract

In adversarial machine learning, deep neural networks can fit the adversarial examples on the training dataset but have poor generalization ability on the test set. This phenomenon is called robust overfitting, and it can be observed when adversarially training neural nets on common datasets, including SVHN, CIFAR-10, CIFAR-100, and ImageNet. In this paper, we study the robust overfitting issue of adversarial training by using tools from uniform stability. One major challenge is that the outer function (as a maximization of the inner function) is nonsmooth, so the standard technique (e.g., hardt et al., 2016) cannot be applied. Our approach is to consider $\\eta$-approximate smoothness: we show that the outer function satisfies this modified smoothness assumption with $\\eta$ being a constant related to the adversarial perturbation $\\epsilon$. Based on this, we derive stability-based generalization bounds for stochastic gradient descent (SGD) on the general class of $\\eta$-approximate smooth functions, which covers the adversarial loss. Our results suggest that robust test accuracy decreases in $\\epsilon$ when $T$ is large, with a speed between $\\Omega(\\epsilon\\sqrt{T})$ and $\\mathcal{O}(\\epsilon T)$. This phenomenon is also observed in practice. Additionally, we show that a few popular techniques for adversarial training (e.g., early stopping, cyclic learning rate, and stochastic weight averaging) are stability-promoting in theory.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2210.00960>)

[PDF](<https://arxiv.org/pdf/2210.00960.pdf>)

[Implementation](<https://github.com/JiancongXiao/Stability-of-Adversarial-Training>)
