Negative Selection Algorithm In Artificial Immune System For Spam Detection

Abstract

Artificial immune system creates techniques that aim at developing immune based models. This was done by distinguishing self from non-self. Mathematical analysis exposed the computation and experimental description of the method and how it is applied to spam detection. This paper looked at evaluation and accuracy in spam detection within the negative selection algorithm. Preliminary result or classifier of self and non-self was carefully studied against mistake of assumption during email classification whereby an email was recognized as a spam and deleted or non-spam and accepted carelessly. This process is called false positive and false negative. Given a threshold, the accuracy increase with increased threshold to determine best performance of the spam detector. Also an improvement of the false positive rate was determined for better spam detector.