A.A. Sirota, M.A. Dryuchenko, E.Yu. Mitrofanova

On the basis of neural network models of functional data conversion is considered an approach for creating gidden digital watermarks in digital content objects based on neural network models, functional data conversion. To work with the data of real format defined general model of information processing of dataembedding. With the use of neural networks back propagation algorithm described steganographic embedding and decoding digital watermarks. For container files with integer kata format with a limited word length proposed modification algorithms for digital watermarking. We investigate the stability of the proposed steganographic algorrithms, and examples of their implementation in the processing of random fields with a given correlation function, as well as handling real objects of digital content (images in formats *.bmp, *.jpeg, audio data in the forman *.wav).