Miao Zhang

PRISM boosts performance in modality-deficient federated graph learning by intelligently retrieving and integrating missing modalities from the entire federation.

Forget catastrophic forgetting: HyLoVQA's hypernetwork-generated LoRA adapters dynamically adapt to new VQA tasks while preserving past knowledge, outperforming prior state-of-the-art methods.

Concept erasure in text-to-image models is mostly smoke and mirrors: a text-free attack can still regenerate "forgotten" concepts by exploiting the model's latent visual knowledge.

Achieve better video editing without retraining by dynamically locking background features based on a "hallucination metric" that detects when the diffusion model is about to go astray.

Multi-agent systems get a 6.3% accuracy boost on math problems thanks to a new "rectify-or-reject" pruning method that dynamically filters out bad information at test time.