Yang Hua

FedMomentum: Preserving LoRA Training Momentum in Federated Fine-Tuning

POLAR: Policy-based Layerwise Reinforcement Learning for Stealthy Backdoor Attacks in Federated Learning

PCEvolve: Private Contrastive Evolution for Synthetic Dataset Generation via Few-Shot Private Data and Generative APIs

Training Diffusion-based Generative Models with Limited Data

FedL2G: Learning to Guide Local Training in Heterogeneous Federated Learning

Poisoning with A Pill: Circumventing Detection in Federated Learning

Exploring Diffusion Models' Corruption Stage in Few-Shot Fine-tuning and Mitigating with Bayesian Neural Networks