SafeFL generates a synthetic dataset from the global model trajectory and uses loss-based median or clustering rules to detect malicious federated learning clients.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Toward Malicious Clients Detection in Federated Learning
SafeFL generates a synthetic dataset from the global model trajectory and uses loss-based median or clustering rules to detect malicious federated learning clients.