HGNN: A Hybrid Graph Neural Network Based on Transfer Learning for Linguistic Steganalysis

Yıl 2024, , 1138 - 1149, 01.10.2024

Merve Varol Arısoy

https://doi.org/10.35414/akufemubid.1427644

Öz

Steganography, especially in the form of text generation based on secret messages, has become a current research topic. It is more difficult to identify the hidden message when it embedded directly into the text without using a cover text, and it also has a higher embedding capacity. Owing to the high rate of imperceptibility and resistance to steganalysis of this type steganography, it is essential that steganalysis methods, generate better performance. Although the complexity of deep learning models increases the accuracy rate, it also increases the inference time. In this study, a linguistic steganalysis was performed with a lower inference time and a higher accuracy rate. In the developed model, first, differences between non-stega and steganographic texts were modelled by a finetuned Bert using the custom dataset. The disparity information obtained by fine-tuned model was distilled into 3 separate networks, BertGCN, BertGAT and BertGIN, for faster and more accurate inference. Then, these 3 distilled networks were combined through Transfer Learning to form a new model. Experiments demonstrates that the proposed model surpass other methods in terms of the accuracy (a success of 0.9879 at 3.22 bpw on text encoded through SAAC Encoding) and the effectiveness of inference (1.09 second).

Anahtar Kelimeler

BERT, GAT (Graph Attention Network), GCN (Graph Convolutional Networks), GIN (Graph Isomorphism Network), knowledge distillation, linguistic steganalysis, transfer learning.

HGNN: Dilsel Steganaliz için Transfer Öğrenimine Dayalı Hibrit Grafik Sinir Ağı

Öz

Özellikle gizli mesajlara dayalı metin üretimi şeklindeki steganografi güncel bir araştırma konusu haline gelmiştir. Gizli mesajın kapak metni kullanılmadan doğrudan metnin içine gömülmesi durumunda tespit edilmesi daha zor olduğu gibi gömme kapasitesi de daha yüksektir. Bu tip steganografinin algılanamazlık oranının yüksek olması ve steganalize karşı direnci nedeniyle, steganaliz yöntemlerinin daha iyi performans üretmesi önemlidir. Derin öğrenme modellerinin karmaşıklığı doğruluk oranını arttırsa da çıkarım süresini de arttırmaktadır. Bu çalışmada, daha düşük çıkarım süresi ve daha yüksek doğruluk oranıyla dilsel steganaliz gerçekleştirilmiştir. Geliştirilen modelde öncelikle stega olmayan ve steganografik metinler arasındaki farklar, özel veri seti kullanılarak hassas ayarlı Bert tarafından modellendi. İnce ayarlı modelle elde edilen eşitsizlik bilgisi, daha hızlı ve daha doğru çıkarım için BertGCN, BertGAT ve BertGIN olmak üzere 3 ayrı ağa ayrıştırıldı. Daha sonra bu 3 damıtılmış ağ, Transfer Öğrenme yoluyla birleştirildi ve yeni bir model oluşturuldu. Deneyler, önerilen modelin doğruluk (SAAC Kodlama yoluyla kodlanan metinde 3,22 bpw'de 0,9879 başarı) ve çıkarımın etkinliği (1,09 saniye) açısından diğer yöntemleri geride bıraktığını göstermektedir.

Anahtar Kelimeler

BERT, GAT (Graph Attention Network), GCN (Graph Convolutional Networks), GIN (Graph Isomorphism Network), knowledge distillation, linguistic steganalysis, transfer learning.

Kaynakça

• Chen, Z., Huang, L., Miao, H., Yang, W., Meng, P. 2011. Steganalysis against substitution-based linguistic steganography based on context clusters. Computers & Electrical Engineering, 37(6), 1071-1081. https://doi.org/10.1016/j.compeleceng.2011.09.014
• Fang, T., Jaggi, M., Argyraki, K. 2017. Generating steganographic text with LSTMs. arXiv preprint arXiv:1705.10742. https://doi.org/10.48550/arXiv.1705.10742
• Fu, Z., Yu, Q., Wang, F., Ding, C. 2022. HGA: Hierarchical feature extraction with graph and attention mechanism for linguistic steganalysis. IEEE Signal Processing Letters, 29, 1734-1738. https://doi.org/10.1109/LSP.2022.3192534
• Jing, W., Song, X., Di, D., Song, H. 2021. GeoGAT: Graph model based on attention mechanism for geographic text classification. Transactions on Asian and Low-Resource Language Information Processing, 20(5), 1-18. https://doi.org/10.1145/3450626
• Kang, H., Wu, H., Zhang, X. 2020. Generative text steganography based on LSTM network and attention mechanism with keywords. Electronic Imaging, 2020(4), 291-1. https://doi.org/10.2352/ISSN.2470-1173.2020.4.MWSF-291
• Kingma, D.P., Ba, J. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980. https://doi.org/10.48550/arXiv.1412.6980
• Kipf, T.N., Welling, M. 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907. https://doi.org/10.48550/arXiv.1609.02907
• Li, S., & Wang, J., Liu, P. 2022. Detection of generative linguistic steganography based on explicit and latent text word relation mining using deep learning. IEEE Transactions on Dependable and Secure Computing, 20(2),1476-148. https://doi.org/10.1109/TDSC.2021.3062703
• Lin, Y., Meng, Y., Sun, X., Han, Q., Kuang, K., Li, J., Wu, F. 2021. BertGCN: Transductive text classification by combining GCN and BERT. arXiv preprint arXiv:2105.05727. https://doi.org/10.48550/arXiv.2105.05727
• Liu, P., Tian, B., Liu, X., Gu, S., Yan, L., Bullock, L., Zhang, W. 2022. Construction of power fault knowledge graph based on deep learning. Applied Sciences, 12(14), 6993. https://doi.org/10.3390/app12146993
• Meng, P., Hang, L., Chen, Z., Hu, Y., Yang, W. (2010). STBS: A statistical algorithm for steganalysis of translation-based steganography. Information Hiding: 12th International Conference. IH 2010. Calgary, AB, Canada, 208-220.
• Meng, P., Hang, L., Yang, W., Chen, Z., Zheng, H. (2009). Linguistic steganography detection algorithm using statistical language model. Proceedings of the 2009 International Conference on Information Technology and Computer Science. Kiev, Ukraine, 25-26.
• Niu, Y., Wen, J., Zhong, P., Xue, Y. 2019. A hybrid R-BILSTM-C neural network based text steganalysis. IEEE Signal Processing Letters, 26(12), 1907-1911. https://doi.org/10.1109/LSP.2019.2955374
• Peng, W., Zhang, J., Xue, Y., Yang, Z. 2021. Real-time text steganalysis based on multi-stage transfer learning. IEEE Signal Processing Letters, 28, 1510-1514. https://doi.org/10.1109/LSP.2021.3105493
• Rassil, A., Chougrad, H., Zouaki, H. (2020). The importance of local labels distribution and dominance for node classification in graph neural networks. Proceedings of the 2020 19th IEEE International Conference on Machine Learning and Applications (ICMLA). Miami, FL, USA, pp. 1505-1511.
• Shen, J., Heng, J., & Han, J. 2020. Near-imperceptible neural linguistic steganography via self-adjusting arithmetic coding. arXiv preprint arXiv:2010.00677. https://doi.org/10.48550/arXiv.2010.00677
• Velickovic, P., Cucurull, G., Casanova, A., Romero, A., Lio, P., Bengio, Y. 2017. Graph attention networks. stat, 1050(20), 10-48550. https://doi.org/10.48550/arXiv.1710.10903
• Wang, H., Li, F. 2022. A text classification method based on LSTM and graph attention network. Connection Science, 34(1), 2466-2480. https://doi.org/10.1080/09540091.2022.2044605
• Wen, J., Zhou, X., Zhong, P., Xue, Y. 2019. Convolutional neural network based text steganalysis. IEEE Signal Processing Letters, 26(3), 460-464. https://doi.org/10.1109/LSP.2019.2895260
• Wu, H., Yi, B., Ding, F., Feng, G., Zhang, X. 2021. Linguistic steganalysis with graph neural networks. IEEE Signal Processing Letters, 28, 558-562. https://doi.org/10.1109/LSP.2021.3058369
• Xiang, L., Liu, Y., You, H., Ou, C. 2022. Aggregating local and global text features for linguistic steganalysis. IEEE Signal Processing Letters, 29, 1502-1506. https://doi.org/10.1109/LSP.2022.3190781
• Xiang, L., Sun, X., Luo, G., Xia, B. 2014. Linguistic steganalysis using the features derived from synonym frequency. Multimedia Tools and Applications, 71, 1893-1911. https://doi.org/10.1007/s11042-012-1303-4
• Xiang, L., Yu, J., Yang, C., Zeng, D., Shen, X. 2018. A word-embedding-based steganalysis method for linguistic steganography via synonym substitution. IEEE Access, 6,64131-64141. https://doi.org/10.1109/ACCESS.2018.2876935
• Xu, K., Hu, W., Leskovec, J., Jegelka, S. 2018. How powerful are graph neural networks?. arXiv preprint arXiv:1810.00826. https://doi.org/10.48550/arXiv.1810.00826
• Yang, H., Bao, Y., Yang, Z., Liu, S., Huang, Y., Jiao, S. (2020). Linguistic steganalysis via densely connected LSTM with feature pyramid. Proceedings of the 2020 ACM Workshop on Information Hiding and Multimedia Security. 5-10.
• Yang, H., Cao, X. 2010. Linguistic steganalysis based on meta features and immune mechanism. Chinese Journal of Electronics, 19, 661-666. https://doi.org/10.1049/cje.2010.661666
• Yang, J., Yang, Z., Zhang, S., Tu, H., Huang, Y. 2021. SeSy: linguistic steganalysis framework integrating semantic and syntactic features. IEEE Signal Processing Letters, 29, 31-35. https://doi.org/10.1109/LSP.2021.3131807
• Yang, Z., Guo, X., Chen, Z., Huang, Y., Zhang, Y. 2019 (a). RNN-Stega: linguistic steganography based on recurrent neural networks. IEEE Transactions on Information Forensics and Security, 14(5), 1280-1295. https://doi.org/10.1109/TIFS.2018.2871746 (a)
• Yang, Z., Huang, Y., Zhang, Y.J. 2019(b). A fast and efficient text steganalysis method. IEEE Signal Processing Letters,26(4),627-631. https://doi.org/10.1109/LSP.2019.2903902 (b)
• Yang, Z., Wang, K., Li, J., Huang, Y., Zhang, Y.J. 2019(c). TS-RNN: text steganalysis based on recurrent neural networks. IEEE Signal Processing Letters, 26(12), 1743-1747. https://doi.org/10.1109/LSP.2019.2950464 (c)
• Yang, Z., Huang, Y., Zhang, Y.J. 2020. TS-CSW: text steganalysis and hidden capacity estimation based on convolutional sliding windows. Multimedia Tools and Applications, 79, 18293-18316. https://doi.org/10.1007/s11042-019-08345-7
• Yang, Z.L., Zhang, S.Y., Hu, Y.T., Hu, Z.W., Huang, Y.F. 2021. VAEStega: linguistic steganography based on variational auto-encoder. IEEE Transactions on Information Forensics and Security, 16, 880-895. https://doi.org/10.1109/TIFS.2020.3037121
• Yao, L., Mao, C., Luo, Y. (2019). Graph convolutional networks for text classification. Proceedings of the AAAI Conference on Artificial Intelligence. 7370-7377.
• Zhang, L., Ding, J., Xu, Y., Liu, Y., Zhou, S. 2021. Weakly-supervised text classification based on keyword graph. arXiv preprint arXiv:2110.02591. https://doi.org/10.48550/arXiv.2110.02591
• Zhang, Y., Xu, Y., Zhang, Y. 2023. A graph neural network node classification application model with enhanced node association. Applied Sciences, 13(12), 7150. https://doi.org/10.3390/app13127150
• Ziegler, Z., Deng, Y., Rush, A. (2019). Neural Linguistic Steganography. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). Hong Kong, China, 1210-1215.
• Zou, J., Yang, Z., Zhang, S., Rehman, S.U., & Huang, Y. (2020). High-Performance Linguistic Steganalysis, Capacity Estimation and Steganographic Positioning. In Digital Forensics and Watermarking: 19th International Workshop, IWDW 2020. Melbourne, VIC, Australia, 80-93.