FedERFT: Improving federated learning through feature-enriched regularization and post-aggregation fine-tuning

Key Points

• Enhancing federated learning improves model performance, particularly in diversified data settings.
• The proposed method shows a significant reduction in model variance, increasing accuracy by up to 15%.
• Analysis of the model employed feature-enriched regularization and post-aggregation fine-tuning techniques.
• This work highlights the need for improved strategies in collaborative machine learning frameworks.