VALUE APPROXIMATION FUNCTIONS FOR PARTIALLY OBSERVED MARKOV DECISION PROCESSES

Pisareva Natalia Dmitrievna

This paper explores the problem of applying value approximation functions in the context of partially observable Markov decision processes (POMDPs, or POMDPs). POMDPs are a powerful tool for modeling situations in which decision making is based on imperfect information and the possible consequences of those decisions.

Keywords:Markov solutions, POMDP, process modeling