How Secure Computation Can Improve Security

Lecture / Panel
For NYU Community

Speaker: Abhi Shelat, University of Virginia

Today's internet systems store personal data about their users to provide their unique utility to their communities. These systems face a constant barrage of attacks, and most are likely to eventually succumb to data breach.

One strategy to mitigate the devastating effects of data breaches is to *limit* the amount of information internet systems collect about their users.  However naive ways to limit data also limit functionality that can be offered.

In this talk, we present our work on practical secure two-party computation as a method to mitigate the risks of data collection while *maintaining* the functionality of candidate internet systems.  Our algorithmic and implementation contributions to secure computation have lead to systems that can process billion gate circuits at the rate of millions of gates per second, a metric that represents improvements over prior methods by factors of $10^{5}$ and $10^4$ respectively. We survey systems that we have developed as well as applications to routing problems, graph problems,  economic stable matching applications, bank ledger audits, and anonymous survey systems.

Contact Nasir Memon for more information.