Big Data Privacy by Design Computation Platform

摘要

We live in the age of Big Data, and personal user data, in particular, is necessary for the operation and improvement of healthcare services. Many times, the capture and use of personal data are not made explicit to the users, but they are central to the business model of companies. However, each person's right to privacy needs to be respected. With the goal of reconciling these two conflicting needs, we designed and implemented a proof-of-concept platform for performing privacy-preserving computations. In particular, we implemented privacy-preserving versions of Machine Learning algorithms, namely Decision Trees, k-Means, Logistic Regression, and Support Vector Machines, using Secure Multi-party Computations with Homomorphic Encryption and Garbled Circuits. For each combination of Machine Learning algorithms with Secure Multi-party Computation techniques, we present the reasoning behind our choices and their potential consequences in terms of performance. The ultimate goal is to provide Privacy-Preserving Computation as a Service. With this platform, we wish to contribute to the faster integration of solutions developed by the scientific community in enterprise systems, thus reducing the time required for innovation to reach products used by many people where privacy improvements are urgently needed.