Privacy preserving data classification using inner product encryption
Abstract
In the context of data outsourcing more and more concerns raise about the privacy of user’s data. One solution is to outsource the data in encrypted form. Meanwhile obtaining a service based on machine learning predictions on user data remains very important in real-life situations. This paper presents ways to combine machine learning algorithms and IPE in order to perform classification on encrypted data. The proposed privacy preserving classification schemes allow to keep user’s data encrypted but at the same time revealing to a server classification results on this data. We study the performance of such classification schemes and their information leakage.