EnCAgg: Enhanced Clustering Aggregation for Robust Federated Learning against Dynamic Model Poisoning
arXiv:2605.22506v1 Announce Type: cross Abstract: Federated learning faces increasing threats from model poisoning attacks, which harms its application to improve privacy. Existing defense methods typically rely on fixed thresholds or perform clustering with a fixed number of clusters to distinguish malicious gradients from benign ones. However, these methods are difficult to adapt to dynamic poisoning strategies of malicious clients, and often result in the loss of benign gradients due to the heterogeneity of clients' local datasets. To address these problems, we propose a novel robust aggreg
The increasing adoption of federated learning in privacy-sensitive sectors makes robust defense against dynamic poisoning attacks a critical and timely research area.
Improving the robustness of federated learning protects the integrity and privacy of AI systems, especially as they become more distributed and decentralized, impacting sectors from healthcare to finance.
This research introduces a more adaptive defense mechanism for federated learning, moving beyond fixed thresholds to better handle sophisticated, dynamic model poisoning attacks.
- · Organizations implementing federated learning
- · Privacy-focused AI applications
- · Cybersecurity researchers
- · Malicious actors attempting model poisoning
- · AI systems vulnerable to gradient attacks
Federated learning deployments become more secure and reliable.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG