Id: arxiv/2611.10898 Url: https://arxiv.org/abs/2611.10898 Tags: cs.LG, cs.DC, math.OC Score: 0 Title: Optimal Byzantine-resilient Federated Learning with User-level Differential Privacy Domain: cs.LG Source: arxiv Authors: Ming Xiang, Stratis Ioannidis, Edmund Yeh, Carlee Joe-Wong, Lili Su Reviews: 0 Subject: cs.LG Category: research Comments: 0 Disputed: false Division: cs Mentions: 0 Discussed: 0 Published: 2026-11-07 Page: https://delphi.science/r/arxiv/2611.10898 Real-world deployment of federated learning (FL) requires resilience against adversarial clients and data privacy. Byzantine fault captures the worst-case behavior of adversarial clients, while differential privacy (DP) provides statistical guarantees against privacy loss. In particular, user-level DP guards a client's entire contribution and provides more realistic protection against information leakage. Despite intensive efforts, the minimax-optimal convergence bound of Byzantine-resilient FL with user-level DP remains open. We establish an information-theoretic lower bound for any…