TITLE:  Exploration vs. Exploitation: Reducing Uncertainty in Operational Problems

ABSTRACT:

Motivated by several core operational applications, we introduce a new class of multistage stochastic optimization models that capture a fundamental tradeoff between performing work and making decisions under uncertainty (exploitation) and investing capacity (and time) to reduce the uncertainty in the decision making (exploration/testing). Unlike existing models, in which the exploration-exploitation tradeoffs typically relate to learning the underlying distributions, the models we introduce assume a known probabilistic characterization of the uncertainty, and focus on the tradeoff of learning (or partially learning) the exact realizations.

Focusing on core scheduling models, we derive insightful structural results on the optimal policies that lead to: (i) Low dimensional dynamic programming formulations; (ii) quantification of the value of learning; (iii) surprising results on the optimality of local (myopic) decision rules for when it is optimal to explore (learn). We then generalize some of the results to a general class of stochastic combinatorial optimization models defined over contra-polymatroids.

The talk is based on several papers that are joint work with Chen Atias, Robi Krauthgamer, Tom Magnanti and Yaron Shaposhnik. 

 Bio:

Retsef Levi is the J. Spencer Standish (1945) Professor of Operations Management at the MIT Sloan School of Management. He is a member of the Operations Management Group at MIT Sloan and affiliated with the Operations Research Center and the Computational for Design and Optimization Program. Prior to joining MIT in 2006, he spent a year as a Goldstine Postdoctoral Fellow at the IBM T.J. Watson Research Center.  He received a Bachelor's degree in Mathematics from Tel-Aviv University in 2001, and a PhD in Operations Research from Cornell University in 2005. He spent more than 11 years in the Israeli Defense Forces as an Office in the Intelligence Wing. Levi's current research is focused on the design of analytical data-driven decision support models and tools addressing complex business and system design decisions under uncertainty in areas, such as health and healthcare management, supply chain, procurement and inventory management, revenue management, pricing optimization and logistics. He is interested in the theory underlying these models and algorithms, as well as their computational and organizational applicability in practical settings. Levi is leading several industry-based collaborative research efforts with some of the major academic hospitals in the Boston area, such as Mass General Hospital (MGH), Beth Israel Deaconess Medical Center (BIDMC), Children’s Hospital, and across the US (e.g., Memorial Sloan Kettering Cancer Center, NYC Prebyterian Hospital System and the American Association of Medical Colleges). Levi is the lead PI on an MIT contract with the Federal Drug Administration (FDA) to develop systematic risk management approach to address risk related to economically motivated adulterations of food and drug products manufactured in China. Levi received the NSF Faculty Early Career Development award, the 2008 INFORMS Optimization Prize for Young Researchers and the 2013 Daniel H. Wagner Prize.