FedAPE: Heterogeneous federated learning with attention-guided aggregation and prototype enhancement

Key Points

• Improved accuracy was achieved through prototype enhancement and attention-guided aggregation methods.
• An accuracy increase of 12% was noted compared to traditional federated learning approaches.
• Analysis of aggregation strategies used attention to prioritize data from diverse sources.
• Enhancements may offer significant benefits but further validation in real-world scenarios is necessary.