Abstract:Recent methods with deep neural networks for text steganalysis have succeeded in mining various feature representations. However, a limited number of studies have explicitly analyzed potential security issues of generative text steganography. Furthermore, current text steganalysis approaches lack detailed consideration in the intricate design of deep learning architectures tailored to these challenges. In this article, in order to tackle these problems, we first theoretically and empirically analyze the inevitable embedding distortions of generative text steganography at a semantic and statistical levels. In light of this, we then propose an innovative text steganalysis method based on hierarchical supervised learning and a dual attention mechanism. Concretely, to extract highly effective semantic features, the proposed method involves fine-tuning a BERT extractor through the hierarchical supervised learning that combines signals from multiple softmax classifiers, rather than relying solely on the final one. The mean and standard deviation values in the Gaussian distribution of cover and stego texts are then estimated using an encoder of variational autoencoders and used to capture features representing the statistical distortion of generative text steganography. Subsequently, we introduce a dual attention mechanism that dynamically fuses the semantic and statistical features, thereby creating discriminative feature representations essential for text steganalysis. The experimental results demonstrate that our proposed text steganalysis method surpasses the current state-of-the-art techniques across three distinct text steganalysis scenarios: specific text steganalysis, semi-blind text steganalysis, and blind text steganalysis.

Linguistic Steganalysis Merging Semantic and Statistical Features.

Small-Scale Linguistic Steganalysis for Multi-Concealed Scenarios

Linguistic Steganalysis Via Densely Connected LSTM with Feature Pyramid

Linguistic Steganalysis Via Fusing Multi-Granularity Attentional Text Features

Linguistic Steganalysis by Enhancing and Integrating Local and Global Features.

High-Performance Linguistic Steganalysis, Capacity Estimation and Steganographic Positioning.

State-of-the-art Advances of Deep-learning Linguistic Steganalysis Research

A Hybrid R-BILSTM-C Neural Network Based Text Steganalysis

Linguistic Steganalysis with Graph Neural Networks

SeSy: Linguistic Steganalysis Framework Integrating Semantic and Syntactic Features

TS-CNN: Text Steganalysis from Semantic Space Based on Convolutional Neural Network

Exploiting Language Model for Efficient Linguistic Steganalysis

Linguistic Steganalysis Toward Social Network

Linguistic Steganography: from Symbolic Space to Semantic Space

Enhancing Steganographic Text Extraction: Evaluating the Impact of NLP Models on Accuracy and Semantic Coherence

Text Steganalysis Based on Hierarchical Supervised Learning and Dual Attention Mechanism.

Linguistic Steganalysis Based on Meta Features and Immune Mechanism

Towards Next-Generation Steganalysis: LLMs Unleash the Power of Detecting Steganography

An Effective Linguistic Steganalysis Framework Based on Hierarchical Mutual Learning

LINK: Linguistic Steganalysis Framework with External Knowledge