PASS EVALUATING IN SIMULATED SOCCER DOMAIN USING ANT-MINER ALGORITHM

This paper utilizes Ant-Miner, the first Ant colony algorithm for discovering classification rules, in a simulated soccer domain to enhance teamwork abilities among agents. The method is used to create appropriate rules using collected training data to estimate whether passing the ball to a particular teammate will result in a success or not. The quality of each constructed rule is considered as estimation for a successful pass. The results showed that the constructed rules with this method are more reliable and accurate in comparison to the widely used constructed rules generated by C4.5