Two Tales of Directed Information

Tsachy Weissman
Stanford University

Abstract:
I will begin by introducing Directed Information. The first part of the talk will show how it arises in the characterization and explicit computation of the fundamental limits on reliable communication. In the second part, I will make the case for caring about how to estimate it and describe some approaches to this estimation problem based on universal sequential probability assignments. I’ll conclude by showcasing an experimental result illustrating the efficacy of directed information estimation as a tool for detecting and measuring causal relevance. Much of the talk is based on collaborations with Himanshu Asnani, Jiantao Jiao, Young-Han Kim, Haim Permuter, and Lei Zhao.

Bio:

Tsachy Weissman has served as a professor of Electrical Engineering at Stanford University since 2003. During 2007-2009, he spent the academic years 2007-2009 on leave with the Department of Electrical Engineering at the Technion.  Prior to his faculty position at Stanford, he worked at Hewlett-Packard Laboratories with the Information Theory Group until 2003.

Prof. Weissman'srResearch focuses on Information Theory and Communications, Statistical Signal Processing, the interplay between them, and their applications. He is the inventor of several patents and is involved in a number of companies as researcher or member of the technical board.

Prof. Weissman's honors include an NSF CAREER award, several best paper awards, Horev fellowship for Leaders in Science and Technology, Henry Taub prize for excellence in research, incumbent of the STMicroelectronics Chair in the School of Engineering, and IEEE fellow.