Boosting Adversarial Training With Mitigating Hard Sample Interference

Summary

This paper proposes mitigating hard sample interference (MHSI), a method for adversarial training that reduces the accuracy-robustness trade-off caused by samples near the decision boundary. It combines a weighted adaptive (WA) mechanism that strengthens learning on clean samples with a dynamic calibration (DC) strategy that adjusts hard-sample probability outputs, guided by gradient norm and Hessian analysis. Experiments on CIFAR-10, CIFAR-100, Tiny ImageNet and SVHN report up to a 6.22% robustness gain over the AT baseline under an l-infinity attack.

Related items

• InfoQCLR: A Plug-in Robustification Module via Quantization for Contrastive Learning in Self-Supervised Skeleton-Based Human Action RecognitionSimilar attack · Arxiv (cs.RO + cs.CV security)
• InfoDoes Target Alignment Mean Target Recovery? An Evidence-Ladder Study of Adversarial Claims on Contrastive EncodersSimilar attack · Arxiv (cs.RO + cs.CV security)
• InfoBRANCH: Bypassing Multi-Scanner AI GuardrailsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoDetecting Adversarial Images through Response Profiles of Vision-Language ModelsSimilar attack · Arxiv (cs.RO + cs.CV security)
• InfoGraphRectify: Graph-Based Transfer of Adversarial Example Detectors Across Neural NetworksSimilar attack · Arxiv (cs.RO + cs.CV security)