Paper Title
Binary Particle Swarm Optimization with Craziness Factor for Feature Selection in Classification

Feature selection plays an important role in classification task. Eliminating irrelevant and redundant feature can simplify the classification learning process and improve the classification performance. However, feature selection is a challenging task due to the large search space. The number of possible solutions increases exponentially with respect to the number of features in the dataset. Binary Particle Swarm Optimization (BPSO) has been applied to solve feature selection problem on binary spaces. However, BPSO can prematurely converge on the local optimum solution. To address this issue, CRAZY-BPSO was proposed by adding a craziness operator with predefined craziness probability to maintain the diversity of the particle and prevent the algorithm to converge prematurely. Moreover, 5-nearest neighbor method was used as a classifier to evaluate the proposed algorithm. The proposed methods obtained the best mean fitness value and accuracy compared with other algorithms in nine out of ten datasets. Thus, the proposed algorithm was superior in term of fitness and classification accuracy. Keywords -