Abstract

A deeper game-tree search can yield a higher decision quality in a heuristic minimax algorithm. However, exceptions can occur as a result of pathological nodes, which are considered to exist in all game trees and can cause a deeper game-tree search, resulting in worse play. To reduce the impact of pathological nodes on the search quality, we propose an iterative optimal minimax (IOM) algorithm by optimizing the backup rule of the classic minimax algorithm. The main idea is that calculating the state values of the intermediate nodes involves not only the static evaluation function involved but also a search into the future, where the latter is given a higher weight. We experimentally demonstrated that the proposed IOM algorithm improved game-playing performance compared to the existing algorithms.

Highlights

  • Computer games are referred to as the “fruit flies” of the discipline of artificial intelligence (AI) [1] with broad applications in areas such as robotics [2], control theory [3], social networks [4], etc

  • To solve the local pathological problems that exist in a game tree and to avoid the shortcomings of the error minimizing minimax (EMM) algorithm, we propose the iterative optimal minimax (IOM) algorithm by refining the backup rule of the minimax algorithm

  • The IOM algorithm and the minimax algorithm yield different search results because the evaluation values of nodes B and C are quite different from the backup values they obtain from their children

Read more

Summary

Introduction

Computer games are referred to as the “fruit flies” of the discipline of artificial intelligence (AI) [1] with broad applications in areas such as robotics [2], control theory [3], social networks [4], etc. Statistic heuristic functions will be used to evaluate the state values of the nodes of depth d, while the values of other nodes of depths less than d are computed in accordance with the minimax rule Refinements to this algorithm have been developed to address various concerns, in particular to improve the efficiency. The incomplete exploration of the game tree and inaccuracy in the evaluation of game situations may hinder the decision accuracy of the minimax algorithm All these studies are based on the assumption that deeper searches typically lead to better performance. The authors in [25,26] first showed that, for certain classes of game trees, the decision quality was degraded by searching deeper and backing up the heuristic values using the minimax propagation rule.

Preliminaries
Minimax
Pathological Node Analysis
Emm Algorithm
Main Idea
Algorithm Description and Theoretical Effectiveness
Experiments and Analysis
Findings
Conclusions and Future Work
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call