Handwritten Digit Recognition Using Clonal Selection Principle in the Artificial Immune System

Abstract

          For centuries, immune system has played a significant role in ensuring the survival and the sustainability of a species. Immune system biologically acts as a shield that protects organism from any pathogen caused diseases or infections by inhibiting any malicious pathogen’s activities. This is accomplished by the mechanism of identifying and consecutively destroying the pathogen upon the detection of it by the lymphocyte cells. Hence, this miraculous ability of the immune system has ignited inspirations and novel approaches to the realm of computing and engineering. Based on the immunological principles, a novel computational technique which mimics the natural immune system is being developed – the Artificial Immune Systems (AIS). By taking the concepts and processes that occurs to lymphocyte cells (B-cells and T-cells) as a foundation, the AIS is composed of three main principles which are Negative Selection, Clonal Selection, and Immune Network Theory. In this study, the Clonal Selection principle is implemented as the classifier for the handwritten digit data by virtue of its ability to distinctively distinguish between the self and non-self antigen. The accuracy of the digit classification done by the Clonal Selection principle is used as the major performance indicator in this study. Empirical study with three datasets shows that the Clonal Selection principle exhibits good classification ability. Experimental results reported the average recognition rate of around 83%. Hence, the results reveal that Clonal Selection principle is capable in recognizing handwritten digit accordingly.

Website : http://ir.fsksm.utm.my/350/