Xiao-Shan Gao

Why Do Unlearnable Examples Work: A Novel Perspective of Mutual Information

Towards a Theoretical Understanding of Prompt Engineering: Tractability, Existence, and Generalization

Exploring Recursive Doubt in Large Language Models

BridgePure: Limited Protection Leakage Can Break Black-Box Data Protection

Provable Watermarking for Data Poisoning Attacks

Generalizability of Neural Networks Minimizing Empirical Risk Based on Expressive Power

Provable Robust Overfitting Mitigation in Wasserstein Distributionally Robust Optimization

Red-Teaming Text-to-Image Systems by Rule-based Preference Modeling

Analyzing the Power of Chain of Thought through Memorization Capabilities

Machine Unlearning for Contrastive Learning under Auditing

Information-theoretically Safe Bias Classifier Against Adversarial Attacks