A Policy Improvement Algorithm for Some Classes of Stochastic Games
Bourque, Matthew J.
MetadataShow full item record
Stochastic games generalize Markov decision processes and repeated games. We give a policy improvement algorithm for additive reward, addi- tive transition (ARAT) zero-sum two-player stochastic games for both discounted and average payoffs. The class of ARAT games includes perfect information games.
Markov decision processes
additive reward additive transition