Inspecting the Running Process of Horizontal Federated Learning via Visual Analytics

Quan Li, Xiguang Wei, Huanbin Lin, Yang Liu, Tianjian Chen, Xiaojuan Ma

View presentation:2021-10-29T15:00:00ZGMT-0600Change your timezone on the schedule page
Exemplar figure, described by caption below
HFLens interface: (A) A data loader for selecting cases of interest; (B) The summary and projection view show the statistical information of the dataset and the 2D embedding projection of all pairs of “client id/server – round id”; (C) The HFL overview presents the participating status of all the involved HFL clients during the HFL running process; (D) The pairwise comparison view displays the metric details of each HFL client in selected communication rounds; (E) The contribution ranking view shows the ranking distributions of all the involved HFL clients.
Fast forward

Direct link to video on YouTube:


Federated learning, anomaly detection, contribution assessment, visualization


As a decentralized training approach, horizontal federated learning (HFL) enables distributed clients to collaboratively learn a machine learning model while keeping personal/private information on local devices. Despite the enhanced performance and efficiency of HFL over local training, clues for inspecting the behaviors of the participating clients and the federated model are usually lacking due to the privacy-preserving nature of HFL. Consequently, the users can only conduct a shallow-level analysis of potential abnormal behaviors and have limited means to assess the contributions of individual clients and implement the necessary intervention. Visualization techniques have been introduced to facilitate the HFL process inspection, usually by providing model metrics and evaluation results as a dashboard representation. Although the existing visualization methods allow a simple examination of the HFL model performance, they cannot support the intensive exploration of the HFL process. In this study, strictly following the HFL privacy-preserving protocol, we design an exploratory visual analytics system for the HFL process termed HFLens, which supports comparative visual interpretation at the overview, communication round, and client instance levels. Specifically, the proposed system facilitates the investigation of the overall process involving all clients, the correlation analysis of clients' information in one or different communication round(s), the identification of potential anomalies, and the contribution assessment of each HFL client. Two case studies confirm the efficacy of our system. Experts' feedback suggests that our approach indeed helps in understanding and diagnosing the HFL process better.