Fomichov O. Methods and models of object classification based on artificial immune systems

The dissertation is devoted to development of methods and models of objects classification using artificial immune systems with different ways of learning that will improve the performance and high accuracy of classification and clustering. The analysis methods for classifying objects, which revealed major shortcomings associated with the problems of classification of low speed high precision separation of objects. A method of a competitive target selection can be applied in algorithms with different immune models in the stage of editing a set of clones or antibodies. Usage of this method allows to increase the speed of immune learning stage, and object classification without loss of accuracy. A method for mutation clones comprising using affinities of elder antibodies as a lower allowable value range can increase the immune learning speed without losing the classification accuracy. Proposed use of stimulating antibodies to enhance immune speed training and the definition of the initial cluster centers, which are formed at the stage of the unsupervised learning, or automatic classification. The generalized model of automatic classification using immune models, can perform not only the distribution of objects between the original set of classes, but also the creation the new clusters for objects, that cannot be attributed to any class. Developed hybrid classification methods which use not only immune data model, but also classical methods of classification (kNN and k-means), as well as soft principles algorithms based on fuzzy logic. Experimental studies of the developed methods and models of objects classification, which have shown their high efficiency. The developed methods and models used to solve the classification problem in the test results of gun owners in "The Guardian" (Kharkiv), as well as determining the amount of insurance payment under compulsory insurance of vehicles classification by their characteristics in the company "Kievian insurance house" (Kharkiv).