Compressing bounded-round communication

Time
1:30 – 2:20pm, Tuesday, March 9, 2010
Place
CSE 305
Speaker
Mark Braverman, Microsoft Research - New England

Abstract

In this talk we will present a near-optimal compression scheme for bounded-round randomized 2-party communication protocol. Previously, such a scheme was only known for protocols where the inputs to the parties are independent.

The results yield a new optimal direct sum theorem for bounded-round communication. They also reveal a tight connection between the Information Cost of a problem and its amortized Communication Complexity.

Joint work with Anup Rao. A talk by Anup on the prequel to this work can be found here:

http://www.cs.washington.edu/education/courses/590z/10wi/media/10wi01.mov