An Improved Firefly Algorithm for Feature Selection in Classification

Feature selection functions as an important method of receiving data so as to make the amount of features decrease. While solving the issue of classifying there exists numerous features having no relevance and being unnecessary which have the potential of making classification performance decrease. Firefly algorithm (FA) functions as an efficient method to make computation which is efficient and progressive. Nevertheless, the conventional FA is easily fallen into the local optima which imposes unsatisfactory practice on feature selection. In this research, one proposal was put forward, the firefly algorithm that combines the binary firefly algorithm with opposition-based learning to select features in classification. Experiment outcomes indicate the fact that the means put forward surpasses PSO and the conventional firefly algorithm.