计算机工程与应用2024，Vol.60Issue(8)：140-147,8.DOI:10.3778/j.issn.1002-8331.2212-0114
尺度不变的条件数约束的模型鲁棒性增强算法
Model Robustness Enhancement Algorithm with Scale Invariant Condition Number Constraint
摘要
Abstract
Deep neural networks are vulnerable to adversarial examples,which has been threatening their application in safety-critical scenarios.Based on the explanation that adversarial examples arise from the highly linear behavior of neural networks,a model robustness enhancement algorithm based on scale-invariant condition number constraint is proposed.Firstly,all weight matrices are used to calculate their norms during the adversarial training process,and the scale-invariant constraint term is obtained through the logarithmic function.Secondly,the scale-invariant condition number constraint item is incorporated into the outer framework of adversarial training optimization,and the condition number value of all weight matrices are iteratively reduced through backpropagation,thereby performing linear transfor-mation of the neural network in a well-conditioned high-dimensional weight space,to improve robustness against adver-sarial perturbations.This algorithm is suitable for visual models of both convolution and Transformer architectures.It can not only significantly improve the robust accuracy against white-box attacks such as PGD and AutoAttack,but also effec-tively enhance the adversarial robustness of defending against black-box attack algorithms including square attack.Incor-porating the proposed constraint during adversarial training on Transformer-based image classification model,the condi-tion number value of weight matrices drops by 20.7%on average,the robust accuracy can be increased by 1.16 percentage points when defending against PGD attacks.Compared with similar methods such as Lipschitz constraints,the pro-posed method can also improve the accuracy of clean examples and alleviate the problem of low generalization caused by adversarial training.关键词
对抗训练/对抗鲁棒性/条件数/尺度不变性/图像分类Key words
adversarial training/adversarial robustness/condition number/scale-invariance/image classification分类
信息技术与安全科学引用本文复制引用
徐杨宇,高宝元,郭杰龙,邵东恒,魏宪..尺度不变的条件数约束的模型鲁棒性增强算法[J].计算机工程与应用,2024,60(8):140-147,8.基金项目
福建省科技计划项目(2021T3003,2021T3068) （2021T3003,2021T3068）
泉州市科技计划项目(2021C065L). （2021C065L）
