English

H+: An Efficient Similarity-Aware Aggregation for Byzantine Resilient Federated Learning

Machine Learning 2025-09-30 v1

Abstract

Federated Learning (FL) enables decentralized model training without sharing raw data. However, it remains vulnerable to Byzantine attacks, which can compromise the aggregation of locally updated parameters at the central server. Similarity-aware aggregation has emerged as an effective strategy to mitigate such attacks by identifying and filtering out malicious clients based on similarity between client model parameters and those derived from clean data, i.e., data that is uncorrupted and trustworthy. However, existing methods adopt this strategy only in FL systems with clean data, making them inapplicable to settings where such data is unavailable. In this paper, we propose H+, a novel similarity-aware aggregation approach that not only outperforms existing methods in scenarios with clean data, but also extends applicability to FL systems without any clean data. Specifically, H+ randomly selects rr-dimensional segments from the pp-dimensional parameter vectors uploaded to the server and applies a similarity check function HH to compare each segment against a reference vector, preserving the most similar client vectors for aggregation. The reference vector is derived either from existing robust algorithms when clean data is unavailable or directly from clean data. Repeating this process KK times enables effective identification of honest clients. Moreover, H+ maintains low computational complexity, with an analytical time complexity of O(KMr)\mathcal{O}(KMr), where MM is the number of clients and KrpKr \ll p. Comprehensive experiments validate H+ as a state-of-the-art (SOTA) method, demonstrating substantial robustness improvements over existing approaches under varying Byzantine attack ratios and multiple types of traditional Byzantine attacks, across all evaluated scenarios and benchmark datasets.

Keywords

Cite

@article{arxiv.2509.24330,
  title  = {H+: An Efficient Similarity-Aware Aggregation for Byzantine Resilient Federated Learning},
  author = {Shiyuan Zuo and Rongfei Fan and Cheng Zhan and Jie Xu and Puning Zhao and Han Hu},
  journal= {arXiv preprint arXiv:2509.24330},
  year   = {2025}
}