Enhancing Contextual Masking in Reversible Linguistic Steganography with Ensemble Methods
M Prasha Meena1, N J S Deepalakshmi2, R Dharsni3, R Subashree4
1Mrs. M. Prasha Meena, Department of Information Technology, Mepco Schlenk Engineering College, Sivakasi (Tamil Nadu), India.
2N J S Deepalakshmi, Department of Information Technology, Mepco Schlenk Engineering College, Sivakasi (Tamil Nadu), India.
3R Dharsni, Department of Information Technology, Mepco Schlenk Engineering College, Sivakasi (Tamil Nadu), India.
4R Subashree, Department of Information Technology, Mepco Schlenk Engineering College, Sivakasi (Tamil Nadu), India.
Manuscript received on 25 April 2024 | Revised Manuscript received on 05 May 2024 | Manuscript Accepted on 15 May 2024 | Manuscript published on 30 May 2024 | PP: 31-40 | Volume-13 Issue-1, May 2024 | Retrieval Number: 100.1/ijrte.A806613010524 | DOI: 10.35940/ijrte.A8066.13010524
Open Access | Editorial and Publishing Policies | Cite | Zenodo | OJS | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Various cybercrimes can be prevented by text authentication, which is responsible for preserving digital identities and contents. Digital signatures are a convenient method for authenticating texts, which is an approach that is widely used. One approach to this problem is linguistic steganography, which enables the hiding of the signature within other words within the text, thereby facilitating efficient data management. However, it should be noted that there is a danger that these kinds of changes may result in inappropriate decisions being made by automated computing systems, not to mention altering their final outputs (unseen). As such, many people are becoming increasingly concerned about the possibility of reversing steganography, making it possible to eliminate any distortions introduced during the process. This paper employs contextual masking instead of random masking with the BERT model. The goal of this research was to develop a natural language text-specific reversible steganographic system. Our model utilises pre-trained BERT as a transformer-based masked language model and reversibly embeds messages through predictive word substitution. To quantify predictive uncertainty, we introduce an adaptive steganographic technique using Bayesian deep learning. This experiment demonstrates how our proposed system strikes a balance between imperceptibility and capacity, while maintaining near-semantic accuracy at all times. Additionally, we employ ensemble methods instead of Monte Carlo to enhance the imperceptibility.
Keywords: Contextual, Ensemble Methods, Reversibility, Steganography.
Scope of the Article: Deep Learning