RESEARCH THE MODEL OF DETECTION POLYMORPHIC MALWARE BY THE CONVOLUTIONAL NEURAL NETWORK

The paper presents the results of research on the use of a convolutional neural network to detect polymorphic malware. The research was conducted on on basis of polymorphic software abc, cheeba, december_3, stasi, otario, dm, v-sign, tequila, flip. The generated of datasets for training a convolutional neural network was carried out using «state matrices» of various dimensions. The Fadeev-Leverrier method was used as a mathematical apparatus. The simulation of the developed software at different iterations and visualization of the results was carried out.