Competition against Fallible Opponents.
Andrew Reibman, Bruce W. Ballard · 1982
Most previous research on search for game playing has focused on improving search efficiency rather than on better utilizing available information. By developing models based on a notion we call playing strength, we acquire the insight needed to develop strategies which perform better than minimax against both perfect and imperfect opponents. In particular situations, our decision strategies yield improvements comparable to or exceeding those provided by an additional ply of search. (Author)