LSHFed: Robust and Communication-Efficient Federated Learning with Locally-Sensitive Hashing Gradient Mapping
Guanjie Cheng, Mengzhen Yang, Xinkui Zhao, Shuyi Yu, Tianyu Du, Yangyang Wu, Mengying Zhu, Shuiguang Deng
TL;DR
LSHFed tackles the dual challenges of robustness and privacy in federated learning by introducing LSHGM, a multi-hyperplane locally-sensitive hashing mechanism that maps gradients to irreversible bit-strings for efficient malicious-gradient detection. Complemented by a mask-based privacy scheme and the ScoreQ-Hash role-election protocol, it enables secure, low-overhead aggregation and dynamic role assignment. Empirical results across MNIST, CIFAR-10, and F-MNIST show LSHFed maintains high accuracy even with up to 50% collusive adversaries and dramatically reduces gradient verification communication (up to 1000x relative to full gradients). The framework thus offers practical, scalable protection against poisoning and privacy leakage while minimizing communication and computation overhead.
Abstract
Federated learning (FL) enables collaborative model training across distributed nodes without exposing raw data, but its decentralized nature makes it vulnerable in trust-deficient environments. Inference attacks may recover sensitive information from gradient updates, while poisoning attacks can degrade model performance or induce malicious behaviors. Existing defenses often suffer from high communication and computation costs, or limited detection precision. To address these issues, we propose LSHFed, a robust and communication-efficient FL framework that simultaneously enhances aggregation robustness and privacy preservation. At its core, LSHFed incorporates LSHGM, a novel gradient verification mechanism that projects high-dimensional gradients into compact binary representations via multi-hyperplane locally-sensitive hashing. This enables accurate detection and filtering of malicious gradients using only their irreversible hash forms, thus mitigating privacy leakage risks and substantially reducing transmission overhead. Extensive experiments demonstrate that LSHFed maintains high model performance even when up to 50% of participants are collusive adversaries while achieving up to a 1000x reduction in gradient verification communication compared to full-gradient methods.