福建师范大学学报（自然科学版）2026，Vol.42Issue(2)：33-42,10.DOI:10.12046/j.issn.1000-5277.2025050021
基于全局-局部扰动协同的对抗样本增强算法
Adversarial Sample Enhancement Algorithm Based on Global-Local Perturbation Collaboration
摘要
Abstract
To enhance the diversity,local adaptability,and transferability of perturbations in in-put transformation attacks,a dynamic global-local adaptive perturbations DGLAP)framework is pro-posed.This framework integrates three core modules to generate and optimize adversarial perturba-tions.The global block shuffling GBS)module reorganizes input information using cross-scale and random diffusion strategies to extract model-invariant features.The local adaptive perturbation LAP module applies diverse transformations to sensitive image regions based on dynamic region partitio-ning and edge-continuity constraints.Additionally,the dynamic weighted random walk DWRW mechanism adaptively adjusts transformation weights through a stochastic strategy that balances ex-ploration and exploitation.Experimental results on the ImageNet dataset demonstrate that DGLAP out-performs baseline methods in terms of attack success rate on mainstream models,such as ResNet18 and ResNet101,and exhibits superior transferability against adversarially trained models.关键词
对抗样本迁移/对抗攻击/黑盒攻击/计算机视觉/输入变换攻击Key words
adversarial sample transferability/adversarial attacks/black-box attacks/computer vision/input transformation attacks分类
信息技术与安全科学引用本文复制引用
陈润泽,叶锋,黄丽清,卢晨浩,陈家祯,黄光樑..基于全局-局部扰动协同的对抗样本增强算法[J].福建师范大学学报（自然科学版）,2026,42(2):33-42,10.基金项目
福建省教育科学规划2024年教育考试招生重点专项课题(FJJKKS24-28) （FJJKKS24-28）
福厦泉国家自主创新示范区协同创新平台项目(2023-P-003) （2023-P-003）
