Chain Association-based Attacking and Shielding Natural Language Processing Systems

JiaCheng Huang, Long Chen
Proceedings of the 16th Asian Conference on Machine Learning, PMLR 260:905-920, 2025.

Abstract

Association as a gift enables people do not have to mention something in completely straightforward words and allows others to understand what they intend to refer to. In this paper, we propose a chain association-based adversarial attack against natural language processing systems, utilizing the comprehension gap between humans and machines. We first generate a chain association graph for Chinese characters based on the association paradigm for building search space of potential adversarial examples. Then, we introduce an discrete particle swarm optimization algorithm to search for the optimal adversarial examples. We conduct comprehensive experiments and show that advanced natural language processing models and applications, including large language models, are vulnerable to our attack, while humans appear good at understanding the perturbed text. We also explore two methods, including adversarial training and associative graph-based recovery, to shield systems from chain association-based attack. Since a few examples that use some derogatory terms, this paper contains materials that may be offensive or upsetting to some people.

Cite this Paper

BibTeX
@InProceedings{pmlr-v260-huang25c, title = {Chain Association-based Attacking and Shielding Natural Language Processing Systems}, author = {Huang, JiaCheng and Chen, Long}, booktitle = {Proceedings of the 16th Asian Conference on Machine Learning}, pages = {905--920}, year = {2025}, editor = {Nguyen, Vu and Lin, Hsuan-Tien}, volume = {260}, series = {Proceedings of Machine Learning Research}, month = {05--08 Dec}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v260/main/assets/huang25c/huang25c.pdf}, url = {https://proceedings.mlr.press/v260/huang25c.html}, abstract = {Association as a gift enables people do not have to mention something in completely straightforward words and allows others to understand what they intend to refer to. In this paper, we propose a chain association-based adversarial attack against natural language processing systems, utilizing the comprehension gap between humans and machines. We first generate a chain association graph for Chinese characters based on the association paradigm for building search space of potential adversarial examples. Then, we introduce an discrete particle swarm optimization algorithm to search for the optimal adversarial examples. We conduct comprehensive experiments and show that advanced natural language processing models and applications, including large language models, are vulnerable to our attack, while humans appear good at understanding the perturbed text. We also explore two methods, including adversarial training and associative graph-based recovery, to shield systems from chain association-based attack. Since a few examples that use some derogatory terms, this paper contains materials that may be offensive or upsetting to some people.} }
Endnote
%0 Conference Paper %T Chain Association-based Attacking and Shielding Natural Language Processing Systems %A JiaCheng Huang %A Long Chen %B Proceedings of the 16th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Vu Nguyen %E Hsuan-Tien Lin %F pmlr-v260-huang25c %I PMLR %P 905--920 %U https://proceedings.mlr.press/v260/huang25c.html %V 260 %X Association as a gift enables people do not have to mention something in completely straightforward words and allows others to understand what they intend to refer to. In this paper, we propose a chain association-based adversarial attack against natural language processing systems, utilizing the comprehension gap between humans and machines. We first generate a chain association graph for Chinese characters based on the association paradigm for building search space of potential adversarial examples. Then, we introduce an discrete particle swarm optimization algorithm to search for the optimal adversarial examples. We conduct comprehensive experiments and show that advanced natural language processing models and applications, including large language models, are vulnerable to our attack, while humans appear good at understanding the perturbed text. We also explore two methods, including adversarial training and associative graph-based recovery, to shield systems from chain association-based attack. Since a few examples that use some derogatory terms, this paper contains materials that may be offensive or upsetting to some people.
APA
Huang, J. & Chen, L.. (2025). Chain Association-based Attacking and Shielding Natural Language Processing Systems. Proceedings of the 16th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 260:905-920 Available from https://proceedings.mlr.press/v260/huang25c.html.

Related Material