Expectation-Based Versus Potential-Aware Automated Abstraction in Imperfect Information Games: An Experimental Comparison Using

ion-Based Versus Potential-Aware Automated Abstraction in Imperfect Information Games: An Experimental Comparison Using Poker Andrew Gilpin and Tuomas Sandholm Computer Science Department Carnegie Mellon University Pittsburgh, PA, USA d abstraction algorithms for sequential imperfect information games have recently emerged as a key component in developing competitive game theory-based agents. The existing literature has not investigated the relative perof different abstraction algorithms. Instead, agents nstruction has used automated abstraction have only been compared under confounding effects: different granuof abstraction and equilibrium-finding algorithms that fferent accuracies when solving the abstracted game. This paper provides the first systematic evaluation of abstraction algorithms. Two families of algorithms have been proposed. The distinguishing feature is the measure used to evaluate the strategic similarity between game states. One algorithm uses the probability of winn...

• Type: Conference
• Year: 2008
• Where: AAAI
• Authors: Andrew Gilpin, Tuomas Sandholm
• Keywords: AAAI 2008, Abstraction Algorithms, Abstractions, Imperfect Information Games, Intelligent Agents