COUNTERING STEGANALYSIS BASED ON GENERATIVE ADVERSARIAL NETWORKS
A. I. Sergadeeva, D. S. Lavrova, E. B. Aleksandrova Peter the Great St. Petersburg Polytechnic University
Annotation: The paper proposes the generative adversarial network approach to improve the robustness of the steganographic method against modern stegoanalyzers. The approach is based on the joint operation of generative adversarial network, pixel importance map and least significant bit replacement method. The results of experimental studies confirmed the effectiveness of the proposed approach
Keywords: generative adversarial networks, steganography, steganography method, steganalysis, machine learning
Pages 57–65