Improved Negative Selection Algorithm with Application to Email Spam Detection
Email spam is an unsolicited message sent into the mail box; it requires the need of an adaptive spam detection model to eliminate these unwanted messages. The random detector generation in negative selection algorithm (NSA) is improved upon by the implementation of particle swarm optimization (PSO) to generate detectors. In the process of detector generation, a fitness function that calculates the reach-ability distance between the non-spam space and the candidate detector, and use the result of the reach-ability distance to calculate the local density factor among the candidate detector was also implemented. The result of the experiment shows an enhancement of the improved NSA over the standard NSA.
KeywordsNegative selection algorithm Particle swarm optimization Local outlier factor Spam Non-spam
Unable to display preview. Download preview PDF.
- 2.Gao, X.Z., Ovaska, S.J., Wang, X.: Particle Swarm Optimization of detectors in Negative Selection Algorithm. In: 2007 ISIC IEEE International Conference on Systems, Man and Cybernetics, October 7-10, pp. 1236–1242 (2007)Google Scholar
- 3.Sotiropoulos, D.: Artificial Immune System-based Machine Learning Methodologies. PhD thesis, University of Piraeus, Piraeus, Greece (2010)Google Scholar
- 5.Oda, T., White, T.: Increasing the accuracy of a spam-detecting artificial immune system. In: The 2003 Congress on Evolutionary Computation, CEC 2003, December 8-12, vol. 391, pp. 390–396 (2003)Google Scholar
- 10.Balthrop, J., Forrest, S., Glickman, M.R.: Revisiting LISYS: parameters and normal behavior. In: Proceedings of the 2002 Congress on Evolutionary Computing, pp. 1045–1050 (2002)Google Scholar
- 11.Forrest, S., Perelson, A.S.: Self nonself discrimination in computer (1994)Google Scholar
- 14.Hopkins, M., Reeber, E., Forman, G., Jaap, S.: Spam Base Dataset. Hewlett-Packard Labs (1999)Google Scholar