From the jungle of stochastic optimization to Sequential Decision Analytics
by: Warren Powell
Sequential decision problems are problems that consist of decision, information, decision, information, …. while incurring costs/receiving rewards. Sequential decision problems cover an incredibly broad problem class, spanning engineering, the sciences, business, economics, finance, health, transportation, energy and e-commerce. The problems may be discrete dynamic programs, continuous control problems, graph problems, stochastic search, active learning, and multiagent games and applications.
We have a very good handle on modeling deterministic optimization problems, but the universe of problems that involve sequential decisions and information have resisted being expressed in the kind of common canonical framework that has become universal in deterministic optimization.
link to website: http://jungle.princeton.edu/