Split Federated Learning

Researchers propose Target-Oriented Feature Decoupling (TOFD), a unified framework that detects and mitigates poisoning attacks in Split Federated Learning (SFL) through three stages: Target Inference…