iPieter / JLIXUZ_pylos

Artificial intelligence: Pylos

This is a Heuristic player for the game Pylos. It uses a rule engine, Drools, to match human insights in the game to a valid move by the player.

Purpose

This was developed for the lab associated with a course on artificial intelligence.

Strategy

The strategy is defined by two sets of files:

• A weights.txt file with the weight for each rule. This file is located in the java folder.
• A set of .drl files, containing the rules writen by us. These are located in the resources folder of the student project.

These rules are used in Java by launching a KieSession, which is a general 'knowledge' framework used by Drools. Once all rules have fired, they will have resulted in a list of moves and a weight, based on the weights.txt file. The highest move is executed.

Weights

To optimize the weights, a genetic algoritm is used. This is implented in PylosMain.java

Results

When matched with a BestFitPlayer, it achieves scores ranging from 90% up to 98% over 100 games.