Data-Free Continual Learning of Server Models in Model-Heterogeneous Federated learning
This work addresses incremental improvements for federated learning systems dealing with model heterogeneity and continual learning, potentially benefiting privacy-preserving distributed AI applications.
The paper tackles the challenges of data heterogeneity, model heterogeneity, and catastrophic forgetting in federated learning by introducing FedDCL, a framework that uses pre-trained diffusion models to enable data-free continual learning of server models, enhancing generalizability in dynamic settings.
Federated learning (FL) is a distributed learning paradigm across multiple entities while preserving data privacy. However, with the continuous emergence of new data and increasing model diversity, traditional federated learning faces significant challenges, including inherent issues of data heterogeneity, model heterogeneity and catastrophic forgetting, along with new challenge of knowledge misalignment. In this study, we introduce FedDCL, a novel framework designed to enable data-free continual learning of the server model in a model-heterogeneous federated setting. We leverage pre-trained diffusion models to extract lightweight class-specific prototypes, which confer a threefold data-free advantage, enabling: (1) generation of synthetic data for the current task to augment training and counteract non-IID data distributions; (2) exemplar-free generative replay for retaining knowledge from previous tasks; and (3) data-free dynamic knowledge transfer from heterogeneous clients to the server. Experimental results on various datasets demonstrate the effectiveness of FedDCL, showcasing its potential to enhance the generalizability and practical applicability of federated learning in dynamic settings.