THE RESEARCH OF IMMUNE OPERATORS IN THE ARTIFICIAL IMMUNE NETWORK MODEL

Abstract

Topicality. Models and methods that allow intelligent processing and systematization of data are of considerable relevance in the field of information technologies. Among such models, one can single out models that work on the basis of biological principles of computing organization, such as artificial neural networks, genetic algorithms, and artificial immune systems. Today, there are several types of models of artificial immune systems that use specific immune operators that affect the speed of the model and the quality of solving practical problems. The goal of this work is a study of the influence of the organization of the work of the most common immune operators on the speed of work of immune algorithms and the accuracy of decision-making regarding the classification of objects with supervised learning. The object of research are еру immune operators of cloning, mutation and suppression, as well as an assessment of the speed of the process of forming an immune response of a network of antibodies to a population of antigens. The subject of research is the model of an artificial immune network and the aiNET algorithm, immune operators and their influence on the process of forming an immune response for the classification of objects with supervised learning. Results. In this paper, the optimal setting of immune operators is proposed to ensure a high speed of immune response formation by the aiNET algorithm when solving the classification problem. In addition, it is proposed to use target immune objects to speed up the classification of antibodies that have not acquired the state of specificity to the antigens of the training sample. Conclusions. The modified aiNET method of classification with supervised learning is planned to be used in the future to control the behavior of characters in action game applications.