Multi-Criteria Ant Feature Selection Using Fuzzy Classifiers

摘要: One of the most important techniques in data preprocessing for mining is feature selection. Real-world analysis, mining, classification and modeling problems usually involve a large number candidate inputs or features. Less relevant highly correlated features decrease, general, accuracy, enlarge complexity classifier. The goal to find reduced set that reveals best accuracy fuzzy This chapter presents an ant colony optimization (ACO) algorithm selection, which minimizes two objectives: error classification. Two pheromone matrices different heuristics are used each objective. performance method compared other selection methods, revealing similar higher performance.