<ul><li>Client selection is crucial in Federated Learning (FL) to maintain high model performance in the presence of challenges like data quality decompensation, budget constraints, and incentive compatibility.</li><li>A new approach called Shapley-Bid Reputation Optimized Federated Learning (SBRO-FL) is proposed to address these challenges by integrating dynamic bidding, reputation modeling, and cost-aware selection.</li><li>Clients in SBRO-FL submit bids based on perceived data quality, and their contributions are evaluated using Shapley values to measure their impact on the global model.</li><li>Experiments on various datasets demonstrate that SBRO-FL enhances accuracy, convergence speed, and robustness, emphasizing the importance of balancing data reliability, incentive compatibility, and cost efficiency for scalable and reliable FL implementations.</li></ul>

Addressing Data Quality Decompensation in Federated Learning via Dynamic Client Selection

Discover more