Robust hierarchical federated learning with dual-layer filtering mechanism

Hierarchical federated learning (HFL) is an effective “cloud-edge-device” distributed model training framework that protects data privacy. During HFL training, poisoning attacks on local data and transmitted models can affect the accuracy of the global model. Existing methods for defending against u...

Full description

Saved in:
Bibliographic Details
Main Authors: Chaoyi Yang, Wei Liang, Yuxiang Chen, Jiahong Xiao, Dacheng He, Tianxiong Liu, Jin Wang, Amr Tolba
Format: Article
Language:English
Published: SAGE Publishing 2025-03-01
Series:International Journal of Advanced Robotic Systems
Online Access:https://doi.org/10.1177/17298806251322260
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849714784370425856
author Chaoyi Yang
Wei Liang
Yuxiang Chen
Jiahong Xiao
Dacheng He
Tianxiong Liu
Jin Wang
Amr Tolba
author_facet Chaoyi Yang
Wei Liang
Yuxiang Chen
Jiahong Xiao
Dacheng He
Tianxiong Liu
Jin Wang
Amr Tolba
author_sort Chaoyi Yang
collection DOAJ
description Hierarchical federated learning (HFL) is an effective “cloud-edge-device” distributed model training framework that protects data privacy. During HFL training, poisoning attacks on local data and transmitted models can affect the accuracy of the global model. Existing methods for defending against unauthorized attacks primarily rely on single-feature anomaly detection approaches, such as calculating distances or densities between data or model parameters. These methods fail to integrate multiple characteristics and metrics to capture anomalous model updates,thus exhibiting significant limitations in model robustness and accuracy. Therefore, we propose a robust hierarchical federated learning method with a dual-layer filtering mechanism (DF-HFL). This method first uses Kernel density estimation to infer the approximate data distribution of devices, ensuring minimal differences among devices within each cluster. It then calculates the density weight of each cluster and the local weight of each device, comparing the difference between local and global weights. Anomalous weights are filtered through a threshold during aggregation. DF-HFL amplifies the distance between malicious updates and normal updates using the dual-layer filtering mechanism, effectively identifying anomalous weights that do not significantly deviate from the normal weight distribution. This helps in accurately detecting anomalies. Additionally, mean filtering is employed to reduce the impact of anomalous data on the original normal gradients, enhancing system robustness. To demonstrate the effectiveness of the proposed method, experiments were conducted on MNIST, FMNIST, Heart Disease and Bank Market datasets. Compared to existing methods like FedAvg, Krum, Random, and Multi-Krum, the global model accuracy improved by 5.18%, 38.98%, 29.96%, and 6.44%.
format Article
id doaj-art-ffada7bda1b949d0a701a26971b768ec
institution DOAJ
issn 1729-8814
language English
publishDate 2025-03-01
publisher SAGE Publishing
record_format Article
series International Journal of Advanced Robotic Systems
spelling doaj-art-ffada7bda1b949d0a701a26971b768ec2025-08-20T03:13:36ZengSAGE PublishingInternational Journal of Advanced Robotic Systems1729-88142025-03-012210.1177/17298806251322260Robust hierarchical federated learning with dual-layer filtering mechanismChaoyi Yang0Wei Liang1Yuxiang Chen2Jiahong Xiao3Dacheng He4Tianxiong Liu5Jin Wang6Amr Tolba7 Hunan Key Laboratory for Service Computing and Novel Software Technology, Xiangtan, China Hunan Key Laboratory for Service Computing and Novel Software Technology, Xiangtan, China Hunan Key Laboratory for Service Computing and Novel Software Technology, Xiangtan, China Hunan Key Laboratory for Service Computing and Novel Software Technology, Xiangtan, China Hunan Key Laboratory for Service Computing and Novel Software Technology, Xiangtan, China , Changsha, China Hunan Key Laboratory for Service Computing and Novel Software Technology, Xiangtan, China Computer Science Department, Community College, , Riyadh Saudi ArabiaHierarchical federated learning (HFL) is an effective “cloud-edge-device” distributed model training framework that protects data privacy. During HFL training, poisoning attacks on local data and transmitted models can affect the accuracy of the global model. Existing methods for defending against unauthorized attacks primarily rely on single-feature anomaly detection approaches, such as calculating distances or densities between data or model parameters. These methods fail to integrate multiple characteristics and metrics to capture anomalous model updates,thus exhibiting significant limitations in model robustness and accuracy. Therefore, we propose a robust hierarchical federated learning method with a dual-layer filtering mechanism (DF-HFL). This method first uses Kernel density estimation to infer the approximate data distribution of devices, ensuring minimal differences among devices within each cluster. It then calculates the density weight of each cluster and the local weight of each device, comparing the difference between local and global weights. Anomalous weights are filtered through a threshold during aggregation. DF-HFL amplifies the distance between malicious updates and normal updates using the dual-layer filtering mechanism, effectively identifying anomalous weights that do not significantly deviate from the normal weight distribution. This helps in accurately detecting anomalies. Additionally, mean filtering is employed to reduce the impact of anomalous data on the original normal gradients, enhancing system robustness. To demonstrate the effectiveness of the proposed method, experiments were conducted on MNIST, FMNIST, Heart Disease and Bank Market datasets. Compared to existing methods like FedAvg, Krum, Random, and Multi-Krum, the global model accuracy improved by 5.18%, 38.98%, 29.96%, and 6.44%.https://doi.org/10.1177/17298806251322260
spellingShingle Chaoyi Yang
Wei Liang
Yuxiang Chen
Jiahong Xiao
Dacheng He
Tianxiong Liu
Jin Wang
Amr Tolba
Robust hierarchical federated learning with dual-layer filtering mechanism
International Journal of Advanced Robotic Systems
title Robust hierarchical federated learning with dual-layer filtering mechanism
title_full Robust hierarchical federated learning with dual-layer filtering mechanism
title_fullStr Robust hierarchical federated learning with dual-layer filtering mechanism
title_full_unstemmed Robust hierarchical federated learning with dual-layer filtering mechanism
title_short Robust hierarchical federated learning with dual-layer filtering mechanism
title_sort robust hierarchical federated learning with dual layer filtering mechanism
url https://doi.org/10.1177/17298806251322260
work_keys_str_mv AT chaoyiyang robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT weiliang robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT yuxiangchen robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT jiahongxiao robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT dachenghe robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT tianxiongliu robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT jinwang robusthierarchicalfederatedlearningwithduallayerfilteringmechanism
AT amrtolba robusthierarchicalfederatedlearningwithduallayerfilteringmechanism