Feature Subset Selection Using Binary Quantum Particle Swarm Optimization for Spam Detection System

E-mail is efficient and common communication method these days, but flooding spam or unsolicited e-mail messages have become uncontrollable. Most spams benefit from the commercial advertising. From the literature review only the accuracy identify the classification process in spam detection. Infrequently, false positive is identified as measurement methods of detection system. Based on the proposed detection system of this study, for the first time Binary Quantum Particle Swarm Optimization (BQPSO) as feature selection method decrease the number of irrelevant features in order to increase classifier performance and decrease dimensionality that influence reliability of detection system. The Multi-Layer Perceptron (MLP) classifier is applied in this research to detect spam emails based on selected relevant features. The experiments are showed on two datasets, namely Ling Spam and Spam Assassin to indicate BOPS based on MLP classifier not only reduce high dimensionality but achieves the accuracy near to 100% with less false positive rate in spam detection system.