Vertical Federated Principal Component Analysis on Feature-wise Distributed Data

Despite the wide attention to federated learning (FL) in the literature, the existing studies mostly focus on supervised federated learning under the horizontally partitioned local dataset setting. This paper will study the unsupervised FL under the vertically partitioned dataset setting. Accordingly, we propose the vertically dataset partitioned federated principal component analysis (VFedPCA) method, which reduces the dimensionality across the joint datasets over all the clients and extracts the principal component feature information for downstream data analysis. VFedPCA features efficient local computation, communication efficiency, and privacy-preserving. Further, we study two communication topologies. The first is a server-client topology where a semi-trusted server coordinates the federated training, while the second is the fully-decentralized topology which eliminates the requirement of the server by allowing clients to communicate with their neighbors. Extensive experiments conducted on real-world datasets justify the efficacy of VFedPCA under vertical partitioned FL setting.