Huili Wang

Differential Privacy (1)

retrieval-augmented generation (2) differential privacy (1) privacy attack (1) model robustness (1) knowledge distillation (1) multimodal learning (1) gradient aggregation (1) poisoning attack (1) vision-language model (1) privacy protection (1) model update (1) membership inference attack (1) gradient perturbation (1) safety awareness (1) adversarial poisoning (1) gradient poisoning (1) harmful output (1) unsafe content (1) federated learning (1) counterfactual attack (1)