Exploiting Language Model For Efficient Linguistic Steganalysis

@article{Yi2022ExploitingLM,
  title={Exploiting Language Model For Efficient Linguistic Steganalysis},
  author={Biao Yi and Hanzhou Wu and Guorui Feng and Xinpeng Zhang},
  journal={ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year={2022},
  pages={3074-3078}
}
  • Biao Yi, Hanzhou Wu, Xinpeng Zhang
  • Published 26 July 2021
  • Computer Science
  • ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Recent advances in linguistic steganalysis have successively applied CNN, RNN, GNN and other efficient deep models for detecting secret information in generative texts. These methods tend to seek stronger feature extractors to achieve higher steganalysis effects. However, we have found through experiments that there actually exists significant difference between automatically generated stego texts and carrier texts in terms of the conditional probability distribution of individual words. Such… 

Figures and Tables from this paper

Autoregressive Linguistic Steganography Based on BERT and Consistency Coding

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.

Semantic-Preserving Linguistic Steganography by Pivot Translation and Semantic-Aware Bins Coding

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.

General Framework for Reversible Data Hiding in Texts Based on Masked Language Modeling

This paper proposes a general framework to embed secret information into a given cover text, for which the embedded information and the original cover text can be perfectly retrieved from the marked text.

References

SHOWING 1-10 OF 27 REFERENCES

Linguistic Steganalysis With Graph Neural Networks

In the proposed method, texts are translated as directed graphs with the associated information, where nodes denote words and edges show associations between the words, and adopted a globally-shared matrix to record correlation strengths between words so that each text can effectively utilize the global information to obtain the better self-representation.

RNN-Stega: Linguistic Steganography Based on Recurrent Neural Networks

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.

TS-RNN: Text Steganalysis Based on Recurrent Neural Networks

This letter observes that the conditional probability distribution of each word in the automatically generated steganographic texts will be distorted after embedded with hidden information and uses recurrent neural networks to extract feature distribution differences and then classify those features into cover text and stego text categories.

Convolutional Neural Network Based Text Steganalysis

This letter proposes a novel text steganalysis model based on convolutional neural network, which is able to capture complex dependencies and learn feature representations automatically from the texts, and uses a word embedding layer to extract the semantic and syntax feature of words.

VAE-Stega: Linguistic Steganography Based on Variational Auto-Encoder

Experimental results show that the proposed model can greatly improve the imperceptibility of the generated steganographic sentences and thus achieves the state of the art performance.

A Fast and Efficient Text Steganalysis Method

This letter proposed a fast and efficient text steganalysis method that can achieve a high detection accuracy and shows a state-of-the-art performance.

Generative Text Steganography Based on LSTM Network and Attention Mechanism with Keywords

Experiments show that the steganographic text generated by the proposed method is of higher semantic quality and more capable of resisting against steganalysis, which has shown the superiority.

Real-Time Text Steganalysis Based on Multi-Stage Transfer Learning

The experimental results show that the proposed text steganalysis method can outperform previously reported methods in terms of detection accuracy and inference efficiency, and enhance inference efficiency and detection performance simultaneously.

Steganalysis against substitution-based linguistic steganography based on context clusters

An Efficient Linguistic Steganography for Chinese Text

A Chinese linguistic steganography algorithm is presented by utilizing the existing Chinese information processing techniques based on the substitution of synonyms and variant forms of the same word in order to decrease the interaction between the surrounding words and the substituted word.