A Novel Ant Colony Optimization Approach to Feature Selection Based on Fuzzy Entropy

Feature selection is a most important procedure which can affect the performance of pattern recognition systems. Since most feature selection algorithms easily fall into local optimum, a novel ant colony optimization approach to feature selection based on fuzzy entropy is proposed (ACOFE). In the proposed algorithm, fuzzy entropy is adopted as pheromone information for ant colony optimization. In order to verify the proposed approach, datasets in UCI Machine Learning Repository are used to test the performance. Simulation experiment results demonstrate that this approach provides higher classification accuracy.