MAIM: A Novel Hybrid Bio-inspired Algorithm for Classification

摘要

The immune system is arguably one of nature’s most highly adaptive, distributed and self-organising systems. It has the property of being able to recognise anomalies — something that deviates from the common rule. The work herein exploits the immune system’s inherently distributed nature in order to enhance classification efficiency and accuracy over that of a more traditional Artificial Immune System (AIS). To this end a novel hybrid classification algorithm MAIM is proposed, combining AIS with an Island Model Genetic Algorithm (IGA). The work herein presents the key features of the MAIM model and investigates the decisions made in the light of the accuracy and efficiency improvements sought.