Colloquium - Hauskrecht

A partially observable Markov decision process (POMDP) provides an elegant mathematical framework for modeling planning and control problems for which there can be uncertainty about the effects of actions and about the current underlying state. However, the computational complexity of the POMDP planning problem limits severely the applicability of the framework and only very small problems can be solved exactly in practice. In the talk, I will focus on two approaches overcoming this difficulty and leading to more efficient solutions: (1) heuristic approximation methods and (2) exploitation of the additional problem structure. I will present the main ideas of the two approaches, new methods I have designed, and their application to medical therapy planning.

Dr. Hauskrecht received his PhD in Computer Science from MIT in August 1997. After finishing his PhD study he worked as a postdoctoral research fellow in the Computer Science Department at Brown University till August 2000 when he joined the Department of Computer and Information Science at Temple University as an assistant professor. His primary field of research interest is Artificial intelligence, in particular, areas of planning, reasoning and optimization in the presence of uncertainty, machine learning, and applications of AI in medicine and finance. His current research work includes projects in stochastic transportation networks, trading in distributed commodity markets, and in planning treatments for patients with ischemic heart disease.

Hosted by Michael Mozer.
Refreshments will be served immediately following the talk in ECOT 831.