Neural Linguistic Steganography

@inproceedings{Ziegler2019NeuralLS,
  title={Neural Linguistic Steganography},
  author={Zachary M. Ziegler and Yuntian Deng and Alexander M. Rush},
  booktitle={EMNLP},
  year={2019}
}
Whereas traditional cryptography encrypts a secret message into an unintelligible form, steganography conceals that communication is taking place by encoding a secret message into a cover signal. Language is a particularly pragmatic cover signal due to its benign occurrence and independence from any one medium. Traditionally, linguistic steganography systems encode secret messages in existing text via synonym substitution or word order rearrangements. Advances in neural language models enable… 

Figures and Tables from this paper

Near-imperceptible Neural Linguistic Steganography via Self-Adjusting Arithmetic Coding
TLDR
A new linguistic steganography method which encodes secret messages using self-adjusting arithmetic coding based on a neural language model which outperforms the previous state-of-the-art methods on four datasets by 15.3% and 38.9% in terms of bits/word and KL metrics, respectively.
Provably Secure Generative Linguistic Steganography
TLDR
A novel provably secure generative linguistic steganographic method ADG, which recursively embeds secret information by Adaptive Dynamic Grouping of tokens according to their probability given by an offthe-shelf language model is presented.
Public Key Neural Linguistic Steganography
TLDR
This work allows users who have not previously communicated secrets with each other to gain the benefits of steganography and provides a construction that eases the pre-coordination requirements for users while providing similar security measures.
Autoregressive Linguistic Steganography Based on BERT and Consistency Coding
TLDR
A novel autoregressive LS algorithm based on BERT and consistency coding is proposed, which achieves a better trade-off between embedding payload and system security and improves theency of the steganographic text while guaranteeing security, and also increases the embedded payload to a certain extent.
Frustratingly Easy Edit-based Linguistic Steganography with a Masked Language Model
TLDR
The proposed method eliminates painstaking rule construction and has a high payload capacity for an edit-based model, and is shown to be more secure against automatic detection than a generation-based method while offering better control of the security/payload capacity trade-off.
Linguistic Steganography: From Symbolic Space to Semantic Space
TLDR
This letter proposed a novel framework of linguistic semantic steganography based on rejection sampling strategy and utilized controllable text generation model for embedding and semantic classifier for extraction.
Semantic-Preserving Linguistic Steganography by Pivot Translation and Semantic-Aware Bins Coding
TLDR
This paper proposes a novel LS method to modify a given text by pivoting it between two different languages and embed secret data by applying a GLS-like information encoding strategy, enabling a high payload to be embedded while keeping the semantic information unchanged.
MTS-Stega: Linguistic Steganography Based on Multi-Time-Step
TLDR
A multi-time-step-based steganography method is proposed, which integrates multiple time steps to select words that can carry secret information and fit the statistical distribution, thus effectively improving the text quality.
Linguistic Generative Steganography With Enhanced Cognitive-Imperceptibility
TLDR
Experimental results show that the proposed methods can further constrain the semantic expression of the generated steganographic text on the basis of ensuring certain perceptual-imperceptibility and statistical-im perceptibility, so as to enhance its cognitive-imPerceptibility.
Meteor: Cryptographically Secure Steganography for Realistic Distributions
TLDR
This work investigates using generative models as steganographic samplers, as they represent the best known technique for approximating human communication, and designs and implements a new steganography protocol, called Meteor, which is provably indistinguishable from honest model output.
...
...

References

SHOWING 1-10 OF 21 REFERENCES
Generating Steganographic Text with LSTMs
TLDR
A new linguistic stegosystem based on a Long Short-Term Memory (LSTM) neural network is proposed that yields high-quality steganographic text while significantly improving capacity relative to the state-of-the-art.
RNN-Stega: Linguistic Steganography Based on Recurrent Neural Networks
TLDR
A linguistic steganography based on recurrent neural networks, which can automatically generate high-quality text covers on the basis of a secret bitstream that needs to be hidden, and achieves the state-of-the-art performance.
LINGUISTIC STEGANOGRAPHY: SURVEY, ANALYSIS, AND ROBUSTNESS CONCERNS FOR HIDING INFORMATION IN TEXT
TLDR
This paper provides a basic introduction to steganography and steganalysis, with a particular focus on text Steganography, and highlights some of the problems inherent in text steganographic as well as issues with existing solutions, and describes linguistic problems with character-based, lexical, and syntactic approaches.
A Novel Linguistic Steganography Based on Synonym Run-Length Encoding
TLDR
Experimental results have shown that the proposed synonym run-length encoding based linguistic steganographic makes fewer changes on the statistical characteristics of cover texts than other algorithms, and enhances the capability of anti-steganalysis.
Steganalysis of Synonym-Substitution Based Natural Language Watermarking
TLDR
This paper tries to distinguish between watermarked articles and unwatermarked articles using context information, and evaluates the suitability of words for their context, and then the suitable sequence of words leads to the final judgment made by a SVM (support vector machine) classifier.
Model-Based Steganography
TLDR
An information-theoretic method for performing steganography and steganalysis using a statistical model of the cover medium is presented, which achieves a higher embedding efficiency and message capacity than previous methods while remaining secure against first order statistical attacks.
Reversible Natural Language Watermarking Using Synonym Substitution and Arithmetic Coding
TLDR
Experimental results illustrate that the proposed novel reversible natural language watermarking method can extract the watermark successfully and recover the original text losslessly and achieves a high embedding capacity.
Steganography- A Data Hiding Technique
TLDR
An in-depth look at how digital images can be used as a carrier to hide messages is discussed and the performance of some of the steganography tools are analyzed.
Avoiding detection on twitter: embedding strategies for linguistic steganography
TLDR
This work investigates the performance of the prior linguistic steganographic method for tweets, combined with perfect coding, and proposes distortion measures for linguistic Steganography, the first of their kind, and investigates the best embedding strategy for the steganographer.
Information hiding using steganography
TLDR
Steganography is a technique of hiding information in digital media that can be extended to copyright protection for digital media: audio, video, and images.
...
...