Privacy Preserving Distributed K-Means Clustering in Malicious Model Using Zero Knowledge Proof

Preserving Privacy is crucial in distributed environments wherein data mining becomes a collaborative task among participants. Critical applications in distributed environment demand higher level of privacy with lesser overheads. Solutions proposed on the lines of cryptography provide higher level of privacy but poor scalability due to higher overheads. Further, existing cryptography based solutions advocate semi honest adversary model while achieving privacy. Practical scenario demands efficient solutions that are secure against malicious adversary model. As per our literature survey, the existing research lacks any fool proof solution for privacy preserving distributed clustering that is efficient and supports malicious adversary model. In this paper, we propose privacy preserving distributed K-Means clustering using Shamir’s Secret Sharing scheme in malicious adversary model. Our empirical evaluation shows that our approach is efficient in terms of computation and communication cost and support malicious adversary model.