Han Xu

Interpretable Probability Estimation with LLMs via Shapley Reconstruction

Position: Multi-Faceted Studies on Data Poisoning can Advance LLM Development

Certified Robustness for Deep Equilibrium Models via Serialized Random Smoothing

Sharpness-Aware Data Poisoning Attack

Confidence-driven Sampling for Backdoor Attacks

DiffusionShield: A Watermark for Data Copyright Protection against Generative Diffusion Models

On the Generalization of Training-based ChatGPT Detection Methods

Exploring Memorization in Fine-tuned Language Models

FT-SHIELD: A Watermark Against Unauthorized Fine-tuning in Text-to-Image Diffusion Models