Abstract
Natural language processing (NLP) presently has become a new paradigm and enables a variety of applications such as text classification, information retrieval, and natural language generation by leveraging deep learning techniques. However, recent studies have shown that the NLP models based on deep neural network are susceptible to maliciously designed adversarial examples. Therefore, the main challenges lie in improving the robustness and ensuring the security of the system. In this paper, we first introduce common NLP tasks and quality measures on adversarial example. Next, we present a comprehensive review on literature in terms of both adversarial attack and defense methods, based upon the granularity and type. Our work is also the first of its kind to provide a brief overview of the adversarial examples on Chinese texts as a result of the language difference. Finally, we summarize this study by providing directions for future research.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
LeCun, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521(7553), 436–444 (2015)
Szegedy, C., Zaremba, W., Sutskever, I., et al.: Intriguing properties of neural networks. In: Proceedings of the 2nd International Conference on Learning Representations, pp. 1–10 (2014)
Wenqi, W., Lina, W., Benxiao, T., et al.: Towards a robust deep neural network in text domain: a survey. arXiv preprint arXiv:1902.07285 (2019)
Zhang, W., Sheng, Q., Alhazmi, A., et al.: Adversarial attacks on deep-learning models in natural language processing: a survey. ACM Trans. Intell. Syst. Technol. 11(3), 24:1–24:41 (2020)
Kusner, M., Sun, Y., Kolkin, N., et al.: From word embeddings to document distances. In: Proceedings of the International Conference on Machine Learning, Lille, pp. 957–966. ACM (2015)
Gao, J., Lanchantin, J., Soffa, M.L., et al.: Black-box generation of adversarial text sequences to evade deep learning classifiers. In: Proceedings of the 2018 IEEE Security and Privacy Workshops, San Francisco, pp. 50–56. IEEE (2018)
He, X., Lyu, L., Xu, Q., et al.: Model extraction and adversarial transferability, your BERT is vulnerable!. In: Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 2006–2012. NAACL (2021)
Ebrahimi, J., Rao, A., Lowd, D., et al.: HotFlip: white-box adversarial examples for text classification. In: Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, Melbourne, pp. 31–36. ACL (2018)
Gil, Y., Chai, Y., Gorodissky, O., et al.: White-to black: efficient distillation of black-box adversarial attacks. In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Minneapolis, pp. 1373–1379. NAACL (2019)
Ebrahimi, J., Lowd, D., Dou, D.: On adversarial examples for character-level neural machine translation. In: Proceedings of the 27th International Conference on Computational Linguistics, Santa Fe, pp. 653–663. ACM (2018)
Ren, S., Deng, Y., He, K., et al.: Generating natural language adversarial examples through probability weighted word saliency. In: Proceedings of the 57th Conference of the Association for Computational Linguistics, Florence, pp. 1085–1097. ACL (2019)
Jin, D., Jin, Z., Zhou, J.T., et al.: Is BERT really robust? A strong baseline for natural language attack on text classification and entailment. In: Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence, New York, pp. 8018–8025. AAAI (2020)
Emmery, C., Kadar, A., Chrupala, G.: Adversarial stylometry in the wild: transferable lexical substitution attacks on author profiling. In: Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pp. 2388–2402. ACL (2021)
Maheshwary, R., Maheshwary, S., Pudi, V.: A strong baseline for query efficient attacks in a black box setting. In: Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, Punta Cana, pp. 8396–8409. ACL (2021)
Zhang, X., Zhang, J., Chen, Z., et al.: Crafting adversarial examples for neural machine translation. In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics, pp. 1967–1977. ACL (2021)
Zeng, Z., Xiong, D.: An empirical study on adversarial attack on NMT: languages and positions matter. In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics, pp. 454–460. ACL (2021)
Emelin, D., Titov, I., Sennrich, R.: Detecting word sense disambiguation biases in machine translation for model-agnostic adversarial attacks. In: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pp. 7635–7653. ACL (2020)
Cheng, Y., Jiang, L., Macherey, W.: Robust neural machine translation with doubly adversarial inputs. In: Proceedings of the 57th Conference of the Association for Computational Linguistics, Florence, pp. 4324–4333. ACL (2019)
Meng, Z., Wattenhofer, R.: A geometry-inspired attack for generating natural language adversarial examples. In: Proceedings of the 28th International Conference on Computational Linguistics, Barcelona, pp. 6679–6689. ACM (2020)
Lin, J., Zou, J., Ding, N.: Using adversarial attacks to reveal the statistical bias in machine reading comprehension models. In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pp. 333–342. ACL (2021)
Wang, T., Wang, X., Qin, Y., et al.: CAT-Gen: improving robustness in NLP models via controlled adversarial text generation. In: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pp. 5141–5146. ACL (2020)
Jia, R., Liang, P.: Adversarial examples for evaluating reading comprehension systems. In: Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, Copenhagen, pp. 2021–2031. ACL (2017)
Wang, B., Pei, H., Pan, B., et al.: T3: tree autoencoder constrained adversarial text generation for targeted attack. In: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pp. 6134–6150. ACL (2020)
Tan, S., Joty, S.R.: Code-mixing on sesame street: dawn of the adversarial polyglots. In: Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 3596–3616. NAACL (2021)
Belinkov, Y., Bisk, Y.: Synthetic and natural noise both break neural machine translation. In: Proceedings of the 6th International Conference on Learning Representations, Vancouver, pp. 1–13. ACM (2018)
Wang, X., Jin, H., Yang, Y., et al.: Natural language adversarial defense through synonym encoding. In: Proceedings of the Thirty-Senventh Conference on Uncertainty in Artificial Intelligence. AUAI (2021)
Zhou, Y., Zheng, X., Hsieh, C.J., et al.: Defense against synonym substitution-based adversarial attacks via Dirichlet neighborhood ensemble. In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pp. 5482–5492. ACL (2021)
Bao, R., Wang, J., Zhao, H.: Defending pre-trained language models from adversarial word substitution without performance sacrifice. In: Proceedings of the Findings of the Association for Computational Linguistics: ACL/IJCNLP 2021, pp. 3248–3258. ACL (2021)
Si, C., Zhang, Z., Qi, F., et al.: Better robustness by more coverage: adversarial and mixup data augmentation for robust finetuning. In: Proceedings of the Findings of the Association for Computational Linguistics: ACL/IJCNLP 2021, pp. 1569–1576. ACL (2021)
Wang, X., Yang, Y., Deng, Y., et al.: Adversarial training with fast gradient projection method against synonym substitution based text attacks. In: Proceedings of the Thirty-Fifth AAAI Conference on Artificial Intelligence, pp. 13997–14005. AAAI (2021)
Mozes, M., Stenetorp, P., Kleinberg, B., et al.: Frequency-guided word substitutions for detecting textual adversarial examples. In: Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pp. 171–186. EACL (2021)
Keller, Y., Mackensen, J., Eger, S.: BERT-defense: a probabilistic model based on BERT to combat cognitively inspired orthographic adversarial attacks. In: Proceedings of the Findings of the Association for Computational Linguistics: ACL/IJCNLP 2021, pp. 1616–1629. ACL (2021)
Wang, B., Wang, S., Cheng, Y., et al.: InfoBERT: improving robustness of language models from an information theoretic perspective. In: Proceedings of the 9th International Conference on Learning Representation (2021)
Le, T., Park, N., Lee, D.: A sweet rabbit hole by DARCY: using honeypots to detect universal trigger’s adversarial attacks. In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pp. 3831–3844. ACL (2021)
Wang, W., Tang, P., Lou, J., et al.: Certified robustness to word substitution attack with differential privacy. In: Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 1102–1112. NAACL (2021)
Ye, M., Gong, C., Liu, Q.: SAFER: a structure-free approach for certified robustness to adversarial word substitution. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 3465–3475. ACL (2020)
Xu, K., Shi, Z., Zhang, H., et al.: Automatic perturbation analysis for scalable certified robustness and beyond. In: Proceedings of the Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020. MIT Press (2020)
Wang, W., Wang, R., Wang, L., Tang, B.: Adversarial examples generation approach for tendency classification on Chinese texts. Ruan Jian Xue Bao/J. Softw. 30(8), 2415–2427 (2019)
Cheng, N., Chang, G., Gao, H., et al.: WordChange: adversarial examples generation approach for Chinese text classification. IEEE Access 8, 79561–79572 (2020)
Yeh, J.F., Lu, Y.Y., Lee, C.H., et al.: Chinese word spelling correction based on rule induction. In: Proceedings of the Third CIPS-SIGHAN Joint Conference on Chinese Language Processing, Wuhan, pp. 139–145. ACL (2014)
Li, J., Du, T., Ji, S., et al.: TextShield: robust text classification based on multimodal embedding and neural machine translation. In: Proceedings of the 29th USENIX Security Symposium, pp. 1381–1398. USENIX Association (2020)
Ian, J.G., Jonathon, S., Christian, S.: Expaining and harnessing adversarial examples. In 3rd International Conference on Learning Representations, San Diego. ICLR (2015)
Ilyas, A., Santurkar, S., Tsipras, D., et al.: Adversarial examples are not bugs, they are features. In: Proceedings of the Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, Vancouver, pp. 125–136. MIT Press (2019)
Acknowledgements
The work is supported by the National Natural Science Foundation of China under Grant 61972148.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Dong, H., Dong, J., Yuan, S., Guan, Z. (2023). Adversarial Attack and Defense on Natural Language Processing in Deep Learning: A Survey and Perspective. In: Xu, Y., Yan, H., Teng, H., Cai, J., Li, J. (eds) Machine Learning for Cyber Security. ML4CS 2022. Lecture Notes in Computer Science, vol 13655. Springer, Cham. https://doi.org/10.1007/978-3-031-20096-0_31
Download citation
DOI: https://doi.org/10.1007/978-3-031-20096-0_31
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-20095-3
Online ISBN: 978-3-031-20096-0
eBook Packages: Computer ScienceComputer Science (R0)