RNN-Stega: Linguistic Steganography Based on Recurrent Neural Networks

Linguistic steganography based on text carrier autogeneration technology is a current topic with great promise and unpredictable challenges. Limited by the text automatic generation technology or the corresponding text coding methods, the quality of the steganographic text generated by previous methods is inferior, which makes its imperceptibility unsatisfactory. In this paper, we propose a linguistic steganography based on Recurrent Neural Networks (RNN-Stega), which can automatically generate high-quality text covers on the basis of a secret bitstream that needs to be embedded. We trained our model with a large number of artificially generated samples and obtained a good estimate of the statistical language model. In the text generation process, we propose Fixed-Length Coding (FLC) and Variable-Length Coding (VLC) to encode words based on their conditional probability distribution. We designed several experiments to test the proposed model from the perspectives of information embedding efficiency, information imperceptibility and information hidden capacity. Experimental results show that the proposed model outperforms all the previous related methods and achieves the state of the art performance.