FedMOP：通过动量正交投影在联邦学习中实现增强的隐私与性能
FedMOP: Achieving Enhanced Privacy and Performance in Federated Learning via Momentum Orthogonal Projection
FedMOP 通过正交投影和动量轨迹隐藏，在联邦学习中同时提升隐私与性能，为视觉模型分布式训练提供抗梯度泄露的通用手段。
原文依据 FedMOP employs gradient orthogonal projection to counteract local drift
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论联邦学习（FL）面临一个根本性困境：现有的针对梯度泄露攻击（GLA）的防御措施总是通过噪声注入或梯度裁剪以牺牲模型性能为代价来保护隐私。我们提出了基于动量的正交投影联邦学习（FedMOP），该方法同时实现了强隐私保证和优越的模型性能。关键见解是利用基于初始化的偏移机制在正交维度上操作。为了提升性能，FedMOP 采用梯度正交投影来抵消局部漂移，利用全局统计上下文有效偏移每个客户端的轮次训练初始模型。为了保护隐私，它引入了基于动量的轨迹偏移隐藏，通过私有初始化和随机演化构建信息屏障，使偏移向量本质上不可恢复。这两种机制是协同而非对立的。理论上，我们证明了收敛保持性，并刻画了攻击者面临的计算上不可行的逆问题。在 CIFAR-10/100 和 Tiny-ImageNet 上的大量实验表明，FedMOP 不仅能有效防御最优的 GLA，而且在准确性和收敛速度上均超越了现有的 FL 方法，验证了其在 FL 中联合增强隐私和性能的能力。代码可在 https://github.com/zyl123456aB/FedMOP 获取。
英文原文摘要 1482 字符
Federated Learning (FL) faces a fundamental dilemma: existing defenses against gradient leakage attacks (GLAs) invariably sacrifice model performance for privacy protection through noise injection or gradient clip. We introduce Federated Learning with Momentum-Based Orthogonal Projection (FedMOP), a method that simultaneously achieves strong privacy guarantees and superior model performance. The key insight is to leverage initialization-based offset mechanisms that operate on orthogonal dimensions. For performance enhancement, FedMOP employs gradient orthogonal projection to counteract local drift, effectively offsetting each client's round-training initial model using global statistical context. For privacy protection, it introduces momentum-based trajectory offset hiding, which makes the offset vector inherently unrecoverable by constructing information barriers through private initialization and randomized evolution. These two mechanisms are synergistic rather than antagonistic. Theoretically, we prove convergence preservation and characterize the computationally infeasible inverse problem faced by attackers. Extensive experiments on CIFAR-10/100 and Tiny-ImageNet demonstrate that FedMOP not only defends effectively against state-of-the-art GLAs but also surpasses existing FL methods in both accuracy and convergence speed, validating its ability to jointly enhance privacy and performance in FL. Codes are available at https://github.com/zyl123456aB/FedMOP.