Game strategy generation method of non-complete information

Abstract

The invention relates to a strategy generation method under non-complete information gaming. The method comprises following specific steps: the opponent behavior modeling step that, before gaming is performed, according to a specific rule of the gaming, modeling for each behavior of an opponent is performed; the behavior collection and information prediction step that, in the gaming process, information of each action of the opponent is recorded, and the information is converted into probability distribution of opponent information through a model; the win rate calculation step which is performed according to a prediction and that, according to the probability distribution, a simulation is performed through a monte carlo method to obtain an own win rate; and the step that an expect benefit of each behavior is calculated according to the win rate finally, and the haviro having the highest expect benefit is selected to be taken as a final decision. A new thought is provided for the non-complete information gaming, the method has the advantages of clear logic, simple implementation and accurate calculations of the opponent information and the win rate, the decision-making capability in specific situation can be improved, a decision benefit is maximized, and then industrial production efficiency is improved.

Description

technical field [0001] The invention relates to a method for generating strategies under complete information games, in particular to a method for generating strategies for incomplete information games with perfect memory features. Background technique [0002] Games can be roughly divided into two types: games under complete information and games under incomplete information. The former means that both parties can fully obtain the current information. For example, in the process of playing Go, Chess and Backgammon, both parties can completely obtain the distribution of all chess pieces on the board and the actions of the opponent; the latter means that both parties cannot obtain the overall information. , For example, in a large number of poker games (Texas Hold'em, bridge, etc.), any player cannot directly watch other players' hands during the game. The solution of game problems is an important issue in the field of artificial intelligence. [0003] In the game of comple...

AI Technical Summary

Problems solved by technology

[0005] The present invention aims at the problems that the information of the opponent cannot be determined and the expected value of the decision cannot be calculated in the game of incomplete information, and provides a method of inferring the information of the opponent according to the action of the opponent, thereby calculating the expected value of the final decision, and making the final decision.

Images