A novel rough set attribute reduction based on ant colony optimisation

摘要: Rough Sets RS, the most promising and proven approach for data reduction, has ability to retain essence of it does not expect any domain inputs from an expert. RS based reduction however can attain only local minima, so elaborate search space is wise employ well known artificial intelligence techniques like ant colony optimisation ACO. In this work a novel rough set attribute on ACO, called as NRSACO proposed, which identify global optimal with help mutual information heuristic aid ants. Few improvements were suggested through minimum reducts attained faster fewer ants iterations. Experiments conducted 22 UCI datasets, results shows that our outperformed in convergence time comparable or improved classification accuracies.