Sample efficient multiagent learning in the presence of Markovian agents (Q2637860)

The book presents the PhD findings of the author in the field of multiagent learning. The agents he proposes exhibit a Markovian behavior: once the values of the discrete feature variables from the joint history of play are consulted, the transition to the next action is performed in a Markovian fashion. All the concepts are thoroughly described and accompanied by theoretical analysis and empirical testing. A book suitable for researchers working in multiagent learning and game theory.