Lizhuang Ma

When AI Agents Collude Online: Financial Fraud Risks by Collaborative LLM Agents on Social Platforms

DORAEMON: Decentralized Ontology-aware Reliable Agent with Enhanced Memory Oriented Navigation

Unintended Harmful Knowledge Elicitation Issue in Large Reasoning Models and a RL Solution

CLIP-Map: Structured Matrix Mapping for Parameter-Efficient CLIP Compression

MBMamba: When Memory Buffer Meets Mamba for Structure-Aware Image Deblurring

One RL to See Them All: Visual Triple Unified Reinforcement Learning

Adaptive Identification of Blurred Regions for Accurate Image Deblurring

3DGS IS A VERSATILE REGULATOR: MODULATING UNIVERSAL METRIC-DEPTH REPRESENTATION VIA ANCHOR-BASED GAUSSIAN-SPLATTED MULTIPLICATION

Tired Actor: Fatigue-Informed Character Control

Towards Differential Handling of Various Blur Regions for Accurate Image Deblurring

MesaTask: Towards Task-Driven Tabletop Scene Generation via 3D Spatial Reasoning

SD-VLM: Spatial Measuring and Understanding with Depth-Encoded Vision-Language Models

SaRA: High-Efficient Diffusion Model Fine-tuning with Progressive Sparse Low-Rank Adaptation

MV-Adapter: Multi-view Consistent Image Generation Made Easy

High-fidelity and realtime 3D Gaussian Head Avatars with Expressive and Compact blenshape representations

Derail Yourself: Multi-turn LLM Jailbreak Attack through self-discovered clues

Diffusion Implicit Policy for Unpaired Scene-aware Motion Synthesis

Textual $D$ecomposition then Sub-motion-space $S$cattering for $O$pen-Vocabulary Motion Generation

Real&Synthetic Dataset and the Linear Attention in Image Restoration