Kejia Chen

Safety at One Shot: Patching Fine-Tuned LLMs with A Single Instance

Multi‑view Adaptive Partitioning with Global Association for Graph Anomaly Detection

CASPO: Confidence-aware Step-wise Preference Optimization for Reliable Reasoning in Large Language Models

Domain-Aware Gradient Reuse for Anomaly Detection

Token-level Inference-Time Alignment for Vision-Language Models

It Takes Two: Your GRPO Is Secretly DPO

Assessing Safety Risks and Quantization-aware Safety Patching for Quantized Large Language Models

Q-resafe: Assessing Safety Risks and Quantization-aware Safety Patching for Quantized Large Language Models