Polymorphic Malware Analysis Model

Abstract

Thе paper presents the results of research on the use of Kohonen neural network in the analysis of polymorphic malware. The assessment of the quality of training was carried out by the fuzzy logic method. The datasets for training the neural network are based on the source code of the polymorphic malware abc, cheeba, december_3, stasi, otario, dm, v-sign, tequila, flip. Simulation of the Kohonen neural network operation at different iterations and visualization of the results was carried out.