This paper proposes FedSC, a novel Sparse Communication mechanism for Federated Learning. • FedSC is designed to reduce both the size an‎d, crucially, the frequency of model uploads. • It achieves high generalization performance an‎d accuracy even with extremely low communication frequency (1-5 uploads total). • The system is particularly effective for non-IID data, a common scenario in IoT.

10/29/2025 12:00:00 AM

88198

1404/08/12

RESEARCH PAPER REVIEW: SPARSE COMMUNICATION MECHANISM FOR FEDERATED LEARNING IN IOT SYSTEMS Student name: Azal Alomairi Student ID: