Horizontal Federated Learning of Deep Multi-view Fuzzy Clustering
Zijian Chen, Jia Miao, Keyu Wu, Xingchen Hu · 2024
Multi-view clustering has become increasingly important in big data analysis, offering a way to combine different data features and uncover hidden patterns. However, this approach presents a critical challenge, as it requires the consolidation of heterogeneous data sources. In various applications, multi-view data are frequently collected and stored in distributed environments. To tackle this issue, we introduce a Deep Multi-view Fuzzy Clustering model built upon Horizontal Federated Learning (FedDMVFCM). This study utilizes deep learning techniques to extract intricate representations from each view of the data, while incorporating fuzzy clustering to manage the inherent uncertainty and overlapping clusters within the datasets. Furthermore, we develop a horizontal federated clustering framework to construct a global cluster model that ensures privacy preservation. Additionally, We develop an optimization algorithm to efficiently address the clustering issue.