Parallel ant colony algorithm for mining classification rules

Abstract

mining classification rules is presented. In the algorithm, each processor is assigned a class label which indicates the consequent parts of the rules it should discover. A group of ants are allocated on each processor to search for the antecedent part of the rules. The ants select the values of the attributes according to the importance of each attribute to the class, the pheromone, and heuristic information. Experimental results on several benchmark datasets show that our algorithm can discover classification rules faster with significantly better accuracy and less redundancy than other methods including the improved Ant-Miner method and the decision-tree-based C4.5 algorithm.