Prompt-based Personalized Federated Learning for Medical Visual Question Answering
This addresses privacy-preserving and personalized medical VQA for healthcare applications, though it appears incremental as it builds on existing pFL methods with optimizations.
The paper tackles data heterogeneity and privacy concerns in medical visual question answering by developing a prompt-based personalized federated learning method, which achieves effective performance across various heterogeneous medical datasets.
We present a novel prompt-based personalized federated learning (pFL) method to address data heterogeneity and privacy concerns in traditional medical visual question answering (VQA) methods. Specifically, we regard medical datasets from different organs as clients and use pFL to train personalized transformer-based VQA models for each client. To address the high computational complexity of client-to-client communication in previous pFL methods, we propose a succinct information sharing system by introducing prompts that are small learnable parameters. In addition, the proposed method introduces a reliability parameter to prevent the negative effects of low performance and irrelevant clients. Finally, extensive evaluations on various heterogeneous medical datasets attest to the effectiveness of our proposed method.