Rediscovering -Minimax Search

The games research community has devoted little effort to investigating search techniques for stochastic domains. The predominant method used in these domains is based on statistical sampling. When full search is required, Expectimax is often the algorithm of choice. However, Expectimax is a full-width search algorithm. A class of algorithms were developed by Bruce Ballard to improve on Expectimax’s runtime. They allow for cutoffs in trees with chance nodes similar to how Alpha-beta allows for cutoffs in Minimax trees. These algorithms were published in 1983—and then apparently forgotten. This paper “rediscovers” Ballard’s *-Minimax algorithms (Star1 and Star2).