Privacy of outsourced two‐party k‐means clustering
Yunlu Cai, Chunming Tang · Concurrency and Computation Practice and Experience · 2019
Summary Many schemes for privacy‐preserving machine learning have been proposed over the past decade. Often, the entities want to keep the privacy of their data while performing machine learning tasks collaboratively, and institutions or end‐users are with limited computing and storage resources. To overcome these issues and to take benefits of cloud computing, it is possible to outsource the execution of a machine learning task to a computing service while retaining confidentiality of the participant's data. Clustering is one of the commonly used tasks in various machine learning and data mining applications. In this paper, we demonstrate that, by using homomorphic encryption, it is possible to outsource the execution of a two‐party k‐means clustering algorithm to a single cloud server while retaining confidentiality of the test data. To the best of our knowledge, ours is the first reasonable scheme to discuss the two‐party k‐means clustering algorithm to a single cloud server.