Fed-CCE: A Multicriteria Framework for Client Selection in Multimodal Heterogeneous Federated Learning

This paper presents a comprehensive review of frameworks addressing heterogeneous federated learning, emphasizing the importance of multimodality and fusion techniques in this context. We explore the challenges and opportunities presented by diverse data types and client nodes in federated learning environments. Building upon this foundation, we propose a novel client selection framework called Composite Client Evaluation in Federated Learning (Fed-CCE), which leverages Multi-Criteria Decision Making (MCDM) techniques. Fed-CCE evaluates potential clients based on multiple parameters, including data quality, computational capacity, historical performance, and accuracy of the local learning model. This approach aims to optimize client selection in heterogeneous federated learning scenarios, potentially improving overall model performance and efficiency. Our work contributes to the growing field of federated learning by addressing the critical issue of client heterogeneity and proposing a practical solution for effective client selection in multimodal environments.