| 摘要 |
Allowed actions of a computer-based player in a game are limited. A pre-action value is identified which corresponds to a relative advantage between the computer-based player and a real person player prior to a current action taken by the computer-based player. A post-action value is identified corresponding to the relative advantage between the computer-based player and the real person player after the current action taken by the computer-based player. The current action is limited to be among a set of allowed actions, which keep the post-action value at least as great as a threshold value. The current action of the computer-based player also may be limited to keep the post-action value not more than a tolerable value, where possible. The current action is selected from the set of allowed actions. The threshold value and tolerance value are prescribed, selected or varied over the course of the game.
|
