Linguistic Steganography Based on Adaptive Probability Distribution

Zhou, Xuejing; Peng, Wanli; Yang, Boya; Wen, Juan; Xue, Yiming; Zhong, Ping

doi:10.1109/tdsc.2021.3079957

Cited by 34 publications

(7 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[58] attempted to address the control of semantic expression in steganographic texts generated by neural networks. Moreover, [59] generated stego texts based on the adaptive probability distribution and the generative adversarial network, which focused on eliminating the exposure bias produced due to the discrepancy between training and inference stages. In [60], Adaptive Dynamic Grouping (ADG) was applied by recursively embedding secret information using an off-the-shelf language tool.…”

Section: Discussion and Future Directionsmentioning

confidence: 99%

See 1 more Smart Citation

A Review on Text Steganography Techniques

et al. 2021

View full text Add to dashboard Cite

There has been a persistent requirement for safeguarding documents and the data they contain, either in printed or electronic form. This is because the fabrication and faking of documents is prevalent globally, resulting in significant losses for individuals, societies, and industrial sectors, in addition to national security. Therefore, individuals are concerned about protecting their work and avoiding these unlawful actions. Different techniques, such as steganography, cryptography, and coding, have been deployed to protect valuable information. Steganography is an appropriate method, in which the user is able to conceal a message inside another message (cover media). Most of the research on steganography utilizes cover media, such as videos, images, and sounds. Notably, text steganography is usually not given priority because of the difficulties in identifying redundant bits in a text file. To embed information within a document, its attributes must be changed. These attributes may be non-displayed characters, spaces, resized fonts, or purposeful misspellings scattered throughout the text. However, this would be detectable by an attacker or other third party because of the minor change in the document. To address this issue, it is necessary to change the document in such a manner that the change would not be visible to the eye, but could still be decoded using a computer. In this paper, an overview of existing research in this area is provided. First, we provide basic information about text steganography and its general procedure. Next, three classes of text steganography are explained: statistical and random generation, format-based methodologies, and linguistics. The techniques related to each class are analyzed, and particularly the manner in which a unique strategy is provided for hiding secret data. Furthermore, we review the existing works in the development of approaches and algorithms related to text steganography; this review is not exhaustive, and covers research published from 2016 to 2021. This paper aims to assist fellow researchers by compiling the current methods, challenges, and future directions in this field.

show abstract

Section: Discussion and Future Directionsmentioning

confidence: 99%

“…A linguistic steganographic method was proposed in [59] that can automatically generate the stego text based on an adaptive probability distribution and a generative adversarial network. The proposed method efficiently addressed the exposure bias produced due to the discrepancy between training and inference stages.…”

Section: Linguisticmentioning

confidence: 99%

A Review on Text Steganography Techniques

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Following [44], we evaluate the fluency of our stego texts with PPL, which is defined as the exponential average negative log-likelihood of a token sequence with a pre-trained language model unless otherwise specified. According to its definition, simply speaking, when tokens with high probability are selected during text generation, the resulting PPL will be lower, meaning that the generated text is more fluent.…”

Section: B Evaluation Metricsmentioning

confidence: 99%

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

Yang¹,

Wu²,

Bu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Linguistic steganography (LS) aims to embed secret information into a highly encoded text for covert communication. It can be roughly divided to two main categories, i.e., modification based LS (MLS) and generation based LS (GLS). Unlike MLS that hides secret data by slightly modifying a given text without impairing the meaning of the text, GLS uses a trained language model to directly generate a text carrying secret data. A common disadvantage for MLS methods is that the embedding payload is very low, whose return is well preserving the semantic quality of the text. In contrast, GLS allows the data hider to embed a high payload, which has to pay the high price of uncontrollable semantics. In this paper, we propose 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. Our purpose is to alter the expression of the given text, enabling a high payload to be embedded while keeping the semantic information unchanged. Experimental results have shown that the proposed work not only achieves a high embedding payload, but also shows superior performance in maintaining the semantic consistency and resisting linguistic steganalysis.

show abstract

“…In [9], proposed a unique linguistic steganographic model which is based on generative adversarial network and an adaptive probability distribution and a achieves the goal of hiding secret messages in generated text while retaining excellent security. To effectively combat exposure bias, the Proposed an approach in [10], which is based on the abstract concept of picture components and can be used to create cover images in JPEG, Bitmap, TIFF, and PNG. The suggested approach is, to our knowledge, the first Steganography algorithm that can work with numerous cover picture formats.…”

Section: Literature Reviewmentioning

confidence: 99%

A High Secured Steganalysis Using QVDHC Model

Akki

Gitanjali

Iwendi

et al. 2022

IJRITCC

View full text Add to dashboard Cite

Data compression plays a vital role in data security as it saves memory, transfer speed is high, easy to handle and secure. Mainly the compression techniques are categorized into two types. They are lossless, lossy data compression. The data format will be an audio, image, text or video. The main objective is to save memory of using these techniques is to save memory and to preserve data confidentiality, integrity. In this paper, a hybrid approach was proposed which combines Quotient Value Difference (QVD) with Huffman coding. These two methods are more efficient, simple to implement and provides better security to the data. The secret message is encoded using Huffman coding, while the cover image is compressed using QVD. Then the encoded data is embedded into cover image and transferred over the network to receiver. At the receiver end, the data is decompressed to obtain original message. The proposed method shows high level performance when compared to other existing methods with better quality and minimum error.

show abstract

Linguistic Steganography Based on Adaptive Probability Distribution

Cited by 34 publications

References 40 publications

A Review on Text Steganography Techniques

A Review on Text Steganography Techniques

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

A High Secured Steganalysis Using QVDHC Model

Contact Info

Product

Resources

About