X-FLoRA: Cross-modal Federated Learning with Modality-expert LoRA for Medical VQA

Figure 1: (a) Illustration for problem definition of cross-modal federated learning, highlighting the complementary strengths of each modality in visual question answering tasks. (b) Visualization of limitations of existing domain adaptation methods using Gaussian kernel density estimation, emphasizing the challenges of adapting MRI and CT modalities. (c) Our proposed method for cross-modal federated learning.