Decentralized Federated Learning Cleans Noisy Labels Using Peer Model Majority Voting
Researchers have introduced a decentralized federated learning framework that corrects noisy training labels across peer-to-peer networks through model-level majority voting ...
