Multiparty Computation Research @ TU Eindhoven: Secure Linear Algebra over the Rationals, and MPyC: Rapid-Prototyping MPC Computations in Python

In this talk I will cover two topics. I will present MPyC, an open-source Python framework for rapid prototyping of MPC protocols, which is being developed at TU Eindhoven. Also, I will present a new MPC protocol for securely solving a full-rank linear system over the rational numbers. More precisely, given a square full-rank matrix with integral coefficients, whose elements are secret-shared among multiple parties, the parties’ goal is to compute the inverse (over Q) in secret-shared form, without leaking any information. Finally I will demonstrate the usefulness of MPyC in practice, by means of explaining step-by-step the MPyC-implementation our new protocol for secure linear algebra.
Speaker Details
Dr. Niek J. Bouman is a postdoctoral researcher at TU Eindhoven, the Netherlands, working on privacy-preserving data mining and multiparty computation. In 2016-2017, Niek was a senior researcher at ABN AMRO bank, working on machine learning topics for automatic detection of fraudulent transactions, and in 2014-2016, he was a post-doctoral researcher at EPFL in Switzerland. Nieks holds a Ph.D. degree (’12) in Mathematics (on quantum cryptography) from Universiteit Leiden, the Netherlands, and from 2008 to 2012, he was with the cryptology group of CWI in Amsterdam. He holds a B.Sc. degree (’05) and M.Sc. degree (’07) in Electrical Engineering from Universiteit Twente, Enschede, NL. His research interests include multiparty computation, privacy, information and communication theory, quantum cryptography, machine learning and statistical inference.
Date:
Speakers:
Dr. Niek J. Bouman
Affiliation:
TU Eindhoven