Five-in-a-row Artificial Intelligence Design And Its Implementation

Game playing is one of the main research fields of artificial intelligence, which involvesartificial intelligence reasoning techniques, search methods and decision processes. In1950,Alan Mathison Turing once said: Playing chess is a very abstract activity, and it’s a field ofintelligence that machine can compete with human. Since then, human-machine playingbecomes a hot research of artificial intelligence. People study five-in-a-row for many years.At present, the optimization of game-tree search algorithm is a hot research field onfive-in-a-row human-machine playing.This paper mainly studies on game-tree search algorithm and alpha-beta pruningtechnology. What’s more, this paper improves the game evaluation function and puts forwardthe concept of continuous Red Four search algorithm.This paper designs an improved evaluation function. The traditional evaluation functionjust considers the chess form of single straight. But it may not judge well in somecomplicated situation. The evaluation function designed in this paper puts in the score ofcross combination based on the single straight. The improved evaluation function can avoidthe computer player missing the must-win situation of Three-Three or Three-Four and so on.It can improve the intelligence during the game procession.This paper studies the professional knowledge of five-in-a-row. Based on it, thealgorithm of continuous Red Four search was put forward. In five-in-a-row game playing,Red Four is a very strong attack means. Unless the opponent can make five chesses in a rowat the next step, opponent must choose the only step for blocking. People often usecontinuous Red Four attack to construct the must-win situation Three-Four. The aim ofcontinuous Red Four search algorithm is finding steps of continuous Red Four which can leadto win, and it’s an assist for game-tree search. What’s more, it can search deeper position thangame-tree search and discover the must-win means in time.Apart from designing the human-machine playing algorithm of five-in-a-row, this paperdesigns a system for five-in-a-row human-machine playing. The system includes fourmodules. The system has favorable expandability since every module is independent.