AMF-CFL: Anomaly model filtering based on clustering in federated learning

Key Points

• Improved anomaly detection reduces false positives in federated learning settings, enhancing model reliability.
• Anomaly detection accuracy increased by 25% when paired with clustering techniques in decentralized networks.
• Analysis using model filtering algorithms demonstrated effective data privacy preservation across multiple devices.
• Findings highlight potential for more robust machine learning models that maintain user data confidentiality.