Lihe Zhang
2 papers · Latest:
Machine Learning
Unveiling the Backdoor Mechanism Hidden Behind Catastrophic Overfitting in Fast Adversarial Training
This paper interprets catastrophic overfitting in fast adversarial training as a backdoor mechanism and proposes new mitigation strategies.
2604.24350
Machine LearningMitigating Error Amplification in Fast Adversarial Training
A new method, DDG, mitigates catastrophic overfitting and improves robustness-accuracy in Fast Adversarial Training by dynamic guidance.
2604.24332
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.