סמינר מחלקתי - ירון שפושניק

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, Retsef Levi, 

and Tom Magnanti.

Short Bio:

Yaron Shaposhnik is a PhD candidate in the Operations Research Center at MIT. He received a Bachelor's 

degree in Information Systems Engineering and a Master's degree in Industrial Engineering, both from 

the Technion - Israel Institute of Technology. His research is focused on Stochastic Dynamic Optimization 

problems with Learning, Data Analytics, and Operations Research Applications (primarily in healthcare).