Novel clustered federated learning based on local loss
This addresses privacy and accuracy challenges in federated learning for distributed systems, representing an incremental advancement in clustered federated learning methods.
The paper tackles the problem of evaluating client data distributions in federated learning by proposing LCFL, a novel clustering metric that improves classification accuracy and applicability to non-convex models without prior knowledge of distributions, achieving superior performance over baselines on benchmarks.
This paper proposes LCFL, a novel clustering metric for evaluating clients' data distributions in federated learning. LCFL aligns with federated learning requirements, accurately assessing client-to-client variations in data distribution. It offers advantages over existing clustered federated learning methods, addressing privacy concerns, improving applicability to non-convex models, and providing more accurate classification results. LCFL does not require prior knowledge of clients' data distributions. We provide a rigorous mathematical analysis, demonstrating the correctness and feasibility of our framework. Numerical experiments with neural network instances highlight the superior performance of LCFL over baselines on several clustered federated learning benchmarks.