Boosting Adversarial Training via Fisher-Rao Norm-based Regularization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yin, Xiangyu, Ruan, Wenjie
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914729373466624
author Yin, Xiangyu
Ruan, Wenjie
author_facet Yin, Xiangyu
Ruan, Wenjie
contents Adversarial training is extensively utilized to improve the adversarial robustness of deep neural networks. Yet, mitigating the degradation of standard generalization performance in adversarial-trained models remains an open problem. This paper attempts to resolve this issue through the lens of model complexity. First, We leverage the Fisher-Rao norm, a geometrically invariant metric for model complexity, to establish the non-trivial bounds of the Cross-Entropy Loss-based Rademacher complexity for a ReLU-activated Multi-Layer Perceptron. Then we generalize a complexity-related variable, which is sensitive to the changes in model width and the trade-off factors in adversarial training. Moreover, intensive empirical evidence validates that this variable highly correlates with the generalization gap of Cross-Entropy loss between adversarial-trained and standard-trained models, especially during the initial and final phases of the training process. Building upon this observation, we propose a novel regularization framework, called Logit-Oriented Adversarial Training (LOAT), which can mitigate the trade-off between robustness and accuracy while imposing only a negligible increase in computational overhead. Our extensive experiments demonstrate that the proposed regularization strategy can boost the performance of the prevalent adversarial training algorithms, including PGD-AT, TRADES, TRADES (LSE), MART, and DM-AT, across various network architectures. Our code will be available at https://github.com/TrustAI/LOAT.
format Preprint
id arxiv_https___arxiv_org_abs_2403_17520
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Boosting Adversarial Training via Fisher-Rao Norm-based Regularization
Yin, Xiangyu
Ruan, Wenjie
Machine Learning
Computer Vision and Pattern Recognition
Adversarial training is extensively utilized to improve the adversarial robustness of deep neural networks. Yet, mitigating the degradation of standard generalization performance in adversarial-trained models remains an open problem. This paper attempts to resolve this issue through the lens of model complexity. First, We leverage the Fisher-Rao norm, a geometrically invariant metric for model complexity, to establish the non-trivial bounds of the Cross-Entropy Loss-based Rademacher complexity for a ReLU-activated Multi-Layer Perceptron. Then we generalize a complexity-related variable, which is sensitive to the changes in model width and the trade-off factors in adversarial training. Moreover, intensive empirical evidence validates that this variable highly correlates with the generalization gap of Cross-Entropy loss between adversarial-trained and standard-trained models, especially during the initial and final phases of the training process. Building upon this observation, we propose a novel regularization framework, called Logit-Oriented Adversarial Training (LOAT), which can mitigate the trade-off between robustness and accuracy while imposing only a negligible increase in computational overhead. Our extensive experiments demonstrate that the proposed regularization strategy can boost the performance of the prevalent adversarial training algorithms, including PGD-AT, TRADES, TRADES (LSE), MART, and DM-AT, across various network architectures. Our code will be available at https://github.com/TrustAI/LOAT.
title Boosting Adversarial Training via Fisher-Rao Norm-based Regularization
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.17520