Leo Ducas (Theory Seminar F14)

"Advances on Fully Homomorphic Encryption"
Leo Ducas
(UCSD)
Monday, November 10th, 2014, 2:00 pm
EBU3B, Room 4258
Abstract:
In 2009  Gentry solved, at least theoretically, the long open problem of building Fully Homomorphic Encryption, that is an encryption scheme that allows anyone to perform arbitrary computation on encrypted data. Such primitive, if reasonably efficient would be extremely relevant in the context of secure cloud computation. While, the firsts constructions were astronomically (yet polynomially) inefficient, much progress has been made on the efficiency front. Prototype implementations are appearing, with performances that could be reasonable for simple applications.

This talk will give an overview of the techniques used to build (leveled) fully homomorphic encryption from the lattice problem "learning with errors", starting with the bootstrapping technique of Gentry, and later improvements for efficiency.