Title: An optimizer's approach to stochastic control problems with nonclassical information structure Speaker: Dr. Ankur Kulkarni Time and Date: 10am, 18/11/11 (Friday)
Mode: Skype based
Venue: HoD Conference Room, 1st Floor, Mechanical Engg. Abstract: We present a novel optimization based approach to the stochastic control problems with nonclassical information structures (in particular, for the long open Witsenhausen problem and its variants- which exemplify the deep relationships between communication, control and team decision theory). In our approach, nonclassicality of information structures corresponds to the nonconvexity of an optimization problem. We show that the most succesful approach to the Witsenhausen problem so far (that of Bansal and Basar) can be explained as obtained from a convexifaction of the nonconvex problem. This talk would be of interest to information/communication theorists, control theorists and optimizers. Speaker's brief Bio: Ankur A. Kulkarni received his B.Tech. in Aerospace Engineering from the Indian Institute of Technology, Bombay in 2006, M.S. in General Engineering and Ph.D. in Industrial Engineering from the University of Illinois at Urbana-Champaign (UIUC) in 2008 and 2010 respectively. He is currently a post-doctoral research associated the Coordinated Science Laboratory at UIUC. He is the recipient of the William A. Chittenden award for an Outstanding M.S. graduate in General Engineering at UIUC. His global interests are in a variety of systems problems, game theory and mathematical programming. His current work is in error correcting codes and stochastic control problems with nonclassical information structures.
News Category