An improved artificial immune recognition system

This paper proposed an improved artificial immune recognition system (IAIRS). It takes multi-training that the affinity threshold scalar ATS is adjusted automatically instead of one shot training, and modifies the criterion of evolving memory cells pool that the affinity threshold AT is only calculated over all samples with the class as the training sample. The IAIRS algorithm is demonstrated on a number of benchmark data sets effectively.

I. Introduction

IN 2001 year, Timmis at al. [1] proposed a resource limited artificial immune system (AIRS). Comparison with linear and nonlinear classifiers, it was be demonstrated on a number of benchmark data sets effectively [2], [3]. It has shown to be successful for the area of remote sensing [4]. Its basic idea can be developed to evolve multiplayer neural networks [5].