ON DETERMINISTIC STATIONARY STRATEGIES FOR MARKOV GAMES

On the Use of Non-Stationary Strategies for Solving Two-Player Zero-Sum Markov Games

The main contribution of this paper consists in extending several non-stationary Reinforcement Learning (RL) algorithms and their theoretical guarantees to the case of γdiscounted zero-sum Markov Games (MGs). As in the case of Markov Decision Processes (MDPs), non-stationary algorithms are shown to exhibit better performance bounds compared to their stationary counterparts. The obtained bounds ...

Stochastic Games and Stationary Strategies

Stationary and convergent strategies in Choquet games

If Nonempty has a winning strategy against Empty in the Choquet game on a space, the space is said to be a Choquet space. Such a winning strategy allows Nonempty to consider the entire finite history of previous moves before making each new move; a stationary strategy only permits Nonempty to consider the previous move by Empty. We show that Nonempty has a stationary winning strategy for every ...

Repeated Games with Stationary Bounded Recall Strategies

A great deal of attention has been paid recently to repeated games with bounded complexity. References [3, 5, 63 and others deal with repeated games played by automata. In this case the set of strategies is reduced to the set of those strategies that can be realized by automata. Here we address ourselves to repeated games played by players with bounded recall who do not know the stage in which ...

Strategy Iteration using Non-Deterministic Strategies for Solving Parity Games

This article introduces the idea of non-deterministic strategies for parity games: In a non-deterministic strategy a player restricts himself to some nonempty subset of possible actions at a given node, instead of limiting himself to exactly one action. We show that a strategy-improvement algorithm by by Björklund, Sandberg, and Vorobyov [3] can easily be adapted to the more general setting of ...