Ant Colony Optimization For Discovering Classification Rules

The focus of this paper is to investigate the classification performance of Ant Colony Optimization (ACO). We evaluate ACO algorithm in three public-domain, real-world datasets used to benchmark the performance of classification algorithms. We compare ACO algorithm to an industry standard PART and JRIP algorithms. The results show that accuracy of ACO algorithm is competitive with PART and JRIP on Breast cancer dataset and produces rule set of small size on adult and mushroom datasets.