Wezel, M.C. van; Eck, N.J.P. van - Erasmus University Rotterdam, Econometric Institute - 2005
Learning, Markov Decision Processes, Dynamic Programming, Neural
Networks, Game Playing, Gaming, Othello.
1 Introduction
Many … problems are the Markov decision
processes (MDPs), described in detail in Section 2. Their most important prop-
erty is that …
to be a Markov decision process (MDP). In an MDP both state transitions and
rewards depend solely on the current state …