The model and algorithm of artificial immune system

An artificial immune system (AIS) model and an algorithm of its performance are suggested in the paper. The model is based on an analogy with typical features of biological immune systems and the principles their functioning. The developed AIS-model has been tested in a image recognition problem (recognition of isolated symbols). The AIS-system contains a single type of active elements that can be identified as B-lymphocytes capable to classification of “own–alien” type. The working algorithm includes two stages: the learning stage (without teacher) and the performance stage. All the types of “alien” symbols (which form the populations to be controlled by the AIS-system) are necessary to be recognized at the learning stage. The population control is realized at the performance stage. The system of computation codes are created providing calculations at all stages of AIS-system work. The calculation experiments have been carried out and the results have been compared with those obtained by alternative methods: by application of multilayered artificial neural networks; by the method of principal components; by the support vectors method.