Ke Xu

Closing the Safety Gap: Surgical Concept Erasure in Visual Autoregressive Models

Influence without Confounding: Causal Discovery from Temporal Data with Long-term Carry-over Effects

SafeSearch: Automated Red-Teaming for the Safety of LLM-Based Search Agents

Revisiting Backdoor Attacks on LLMs: A Stealthy and Practical Poisoning Framework via Harmless Inputs

On the Self-awareness of Large Reasoning Models' Capability Boundaries

Understanding the Dilemma of Unlearning for Large Language Models

ALMGuard: Safety Shortcuts and Where to Find Them as Guardrails for Audio–Language Models

FlowRefiner: A Robust Traffic Classification Framework against Label Noise

An Engorgio Prompt Makes Large Language Model Babble on

FacLens: Transferable Probe for Foreseeing Non-Factuality in Large Language Models

One Perturbation is Enough: On Generating Universal Adversarial Perturbations against Vision-Language Pre-training Models

CALoR: Towards Comprehensive Model Inversion Defense

MIBench: A Comprehensive Benchmark for Model Inversion Attack and Defense