Artificial Immune Tissue using Self-Orgamizing Networks

TL;DR

This paper introduces a novel artificial immune tissue component using self-organizing networks, enhancing data pre-processing, clustering, and feature extraction in immune-inspired algorithms.

Contribution

It proposes a new tissue algorithm based on self-organizing maps and Toll-Like Receptor analogies, adding a tissue layer to AIS for improved data processing.

Findings

Enhanced data clustering and feature extraction capabilities

Potential for improved immune-inspired computational algorithms

Integration of TLR analogies improves activation functions

Abstract

As introduced by Bentley et al. (2005), artificial immune systems (AIS) are lacking tissue, which is present in one form or another in all living multi-cellular organisms. Some have argued that this concept in the context of AIS brings little novelty to the already saturated field of the immune inspired computational research. This article aims to show that such a component of an AIS has the potential to bring an advantage to a data processing algorithm in terms of data pre-processing, clustering and extraction of features desired by the immune inspired system. The proposed tissue algorithm is based on self-organizing networks, such as self-organizing maps (SOM) developed by Kohonen (1996) and an analogy of the so called Toll-Like Receptors (TLR) affecting the activation function of the clusters developed by the SOM.