Princeton University

School of Engineering & Applied Science

Secure Multiparty Computation: From Theory to Google

Speaker: 
Benjamin Kreuter, Google
Location: 
Engineering Quadrangle B205
Date/Time: 
Tuesday, November 28, 2017 - 11:00am to 12:00pm

Abstract:  In many Internet applications, computations must be performed on data held by multiple parties.  Often this data is considered private or sensitive, and there are growing concerns about the security risks that sharing such data entails.  One way to address those concerns is to use Secure Multiparty Computation (MPC), cryptographic techniques that allow multiple parties to compute a function over private inputs without having to reveal those inputs.  Although MPC was first described by researchers in the 1980s, real-world applications have only been deployed in the past few years, and practicality remains a challenge in many settings.
 
This talk will describe the progress on practical MPC deployment and the engineering challenges facing practitioners.  Applications deployed at Google will be used to illustrate some of these challenges.  This talk will also describe the role that theoretical results play in designing practical applications.
 
Bio:  Benjamin Kreuter works on MPC applications at Google, and is nearing the completion of his PhD at Northeastern University.  His research interest is in practical multiparty computation and homomorphic encryption, and related topics in cryptography and distributed systems.