Kui Ren

Mitigating Over-Refusal in Adversarial Tuning via Subspace-guided Sample Selection

Towards Real-world Debiasing: Rethinking Evaluation, Challenge, and Solution

Untargeted Jailbreak Attack

Explainable Token-level Noise Filtering for LLM Fine-tuning Datasets

The Robustness-Security Paradox: Channel-Aware Feature Learning for Adversarial Watermark Exploitation

Dynamic Target Attack

Module-Aware Parameter-Efficient Machine Unlearning on Transformers

CoKV: Optimizing LLM Inference with Game-Theoretic Adaptive KV Cache

CARP: Causal Alignment of Reward Models via Response-to-Prompt Prediction

Rethinking Shapley Value for Data Contribution

Towards Evaluation for Real-World LLM Unlearning

CollabMask: Explainable Neuron Collaboration with Gradient Masks for LLM Fine-Tuning

ZeroSecBench: Fine-grained and Robust Evaluation for Secure Code Generation

WMCopier: Forging Invisible Watermarks on Arbitrary Images

Textual Unlearning Gives a False Sense of Unlearning

Taught Well Learned Ill: Towards Distillation-conditional Backdoor Attack

Robust Representation Consistency Model via Contrastive Denoising

JudgeRail: Harnessing Open-Source LLMs for Fast Harmful Text Detection with Judicial Prompting and Logit Rectification

Towards Real World Debiasing: A Fine-grained Analysis On Spurious Correlation

REFINE: Inversion-Free Backdoor Defense via Model Reprogramming

DAG-SHAP: Feature Attribution in DAG based on Edge Intervention

Structure-Aware Parameter-Efficient Machine Unlearning on Transformer Models

Prompt-Consistency Image Generation (PCIG): A Unified Framework Integrating LLMs, Knowledge Graphs, and Controllable Diffusion Models

Towards False-claim-resistant Model Ownership Verification via Targeted Fingerprint

A Comprehensive Deepfake Detector Assessment Platform