<ul data-eligibleForWebStory="true"><li>Ensuring resilience to Byzantine clients while protecting the privacy of data in federated learning is a challenge.</li><li>Existing secure aggregation techniques are effective when clients' data is homogeneous but fail for heterogeneous data.</li><li>Pre-processing techniques like nearest neighbor mixing can enhance countermeasures in heterogeneous settings.</li><li>Proposed multi-stage method combines secret sharing, secure aggregation, and private information retrieval for privacy and resilience.</li><li>The method is designed to provide information-theoretic privacy guarantees and Byzantine resilience under data heterogeneity.</li><li>Scheme outperforms previous techniques in combating various attacks in federated learning.</li><li>Investigation into reducing communication overhead of secure aggregation through zero-order estimation methods.</li><li>Efforts to make private aggregation scalable in state-of-the-art federated learning tasks.</li></ul>

Private Aggregation for Byzantine-Resilient Heterogeneous Federated Learning

Discover more