Yu Zhang

Discrete Latent Features Ablate Adversarial Attack: A Robust Prompt Tuning Framework for VLMs

Semi-Supervised Noise Adaptation: Transferring Knowledge from Noise Domain

MoSA: Mosaic Shared Adaptation of Large Language Models

Reasoning-Aligned Perception Decoupling for Scalable Multi-modal Reasoning

SPARD: Defending Harmful Fine-Tuning Attack via Safety Projection with Relevance–Diversity Data Selection

Heterogeneous-Modal Unsupervised Domain Adaptation via Latent Space Bridging

From Coefficients to Directions: Rethinking Model Merging with Directional Alignment

FABLE: Federated Anchor-Based Learning with Privacy Protection

Heterogeneous Incremental Learning for Dense Prediction: Advancing Knowledge Retention via Self-Distillation

Agentic Robot: A Brain-Inspired Framework for Vision-Language-Action Models in Embodied Agents

One-Token Verification for Reasoning LLMs, Anytime, Anywhere

Equivariant CNNs via Hypernetworks

Federated Unlearning with Gradient Shielding

Merging without Forgetting: Continual Fusion of Task-Specific Models via Optimal Transport

HiRA: Parameter-Efficient Hadamard High-Rank Adaptation for Large Language Models

Sharpness-Aware Black-Box Optimization

MoPFormer: Motion-Primitive Transformer for Wearable-Sensor Activity Recognition

Enhancing Pruned Models by Input Compensation

ComLoRA: A Competitive Learning Approach for Enhancing LoRA

MTSAM: Multi-Task Fine-Tuning for Segment Anything Model

Come Together, But Not Right Now: A Progressive Strategy to Boost Low-Rank Adaptation

Graph Vision Networks for Link Prediction

HeadMap: Locating and Enhancing Knowledge Circuits in LLMs

EMOVA: Empowering Language Models to See, Hear and Speak with Vivid Emotions

Hypernetwork-Based Equivariant CNNs

Input Compensation for Pruned Models

Mixture of insighTful Experts (MoTE): The Synergy of Thought Chains and Expert Mixtures in Safety Self-Alignment

Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction