Privacy-preserving Protocol over Vertically Partitioned Data in Multiparty K-means Clustering

The article examines and analyzes existing solutions of privacy -preserving computations over vertical partitioning in multiparty clustering by k -means. The modifications of cryptographic primitives to eliminate detected deficiencies are offered. The results of comparative analysis of advanced cryptographic primitives are given. The protocol is developed that preserves data privacy with few parties considering their possible collusion. The conclusions on the effectiveness of the improved cryptographic primitives are made and the recommendations how to use the developed protocol are given.