Lei Feng

Hierarchy-of-Groups Policy Optimization for Long-Horizon Agentic Tasks

The Final Layer Holds the Key: A Unified and Efficient GNN Calibration Framework

Mitigating Mismatch within Reference-based Preference Optimization

TokenSwap: Backdoor Attack on the Compositional Understanding of Large Vision-Language Models

Beyond Hard Supervised Fine-tuning: Enhancing Image-text Alignment of Strong Models with Weak Models

Revisiting Confidence Calibration for Misclassification Detection in VLMs

Mitigating Strategy-Selection Bias in Reasoning for More Effective Test-Time Scaling

CARPRT: Class-Aware Zero-Shot Prompt Reweighting for Vision-Language Model

Do All Individual Layers Help? An Empirical Study of Task-Interfering Layers in Vision-Language Models

Reasoning-Preserved Safety Alignment for Large Reasoning Models

Solving Puzzles? Jailbreaking Multimodal Large Language Models!

Improving Generalizability and Undetectability for Targeted Adversarial Attacks on Multimodal Pre-trained Models

Inference-time Unlearning via Adaptive Output Regulation

Rethinking Toxicity Evaluation in Large Language Models: A Multi-Label Perspective

Establishing Linear Surrogate Regret Bounds for Convex Smooth Losses via Convolutional Fenchel–Young Losses

Instance-dependent Early Stopping

Representation Surgery in Model Merging with Probabilistic Modeling

Understanding Model Reprogramming for CLIP via Decoupling Visual Prompts

Rethinking Chain-of-Thought from the Perspective of Self-Training

A Closer Look at Backdoor Attacks on CLIP

Enhancing the Influence of Labels on Unlabeled Nodes in Graph Convolutional Networks

Test-Time Multimodal Backdoor Detection by Contrastive Prompting

Attribute-based Visual Reprogramming for Vision-Language Models

Automatic Dataset Construction (ADC): Sample Collection, Data Curation, and Beyond

Defending Multimodal Backdoored Models by Repulsive Visual Prompt Tuning

BDetCLIP: Multimodal Prompting Contrastive Test-Time Backdoor Detection

Endowing Visual Reprogramming with Adversarial Robustness

Enhancing Sample Selection Against Label Noise by Cutting Mislabeled Easy Examples

A Closer Look at Backdoor Attacks on CLIP

Delving into Temperature Scaling for Adaptive Conformal Prediction

Exploiting Presentative Feature Distributions for Parameter-Efficient Continual Learning of Large Language Models

Dual-Head Knowledge Distillation: Enhancing Logits Utilization with an Auxiliary Head

Prototype-based Optimal Transport for Out-of-Distribution Detection

ELU-GCN: Effectively Label-Utilizing Graph Convolutional Network

Exploring Weak-to-Strong Generalization for CLIP-based Classification