Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systems

Steffen Eger, Gözde Gül Şahin, Andreas Rücklé, Ji-Ung Lee, Claudia Schulz, Mohsen Mesgar, Krishnkant Swarnkar, Edwin Simpson, Iryna Gurevych


Abstract
Visual modifications to text are often used to obfuscate offensive comments in social media (e.g., “!d10t”) or as a writing style (“1337” in “leet speak”), among other scenarios. We consider this as a new type of adversarial attack in NLP, a setting to which humans are very robust, as our experiments with both simple and more difficult visual perturbations demonstrate. We investigate the impact of visual adversarial attacks on current NLP systems on character-, word-, and sentence-level tasks, showing that both neural and non-neural models are, in contrast to humans, extremely sensitive to such attacks, suffering performance decreases of up to 82%. We then explore three shielding methods—visual character embeddings, adversarial training, and rule-based recovery—which substantially improve the robustness of the models. However, the shielding methods still fall behind performances achieved in non-attack scenarios, which demonstrates the difficulty of dealing with visual attacks.
Anthology ID:
N19-1165
Volume:
Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)
Month:
June
Year:
2019
Address:
Minneapolis, Minnesota
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
1634–1647
Language:
URL:
https://aclanthology.org/N19-1165
DOI:
10.18653/v1/N19-1165
Bibkey:
Cite (ACL):
Steffen Eger, Gözde Gül Şahin, Andreas Rücklé, Ji-Ung Lee, Claudia Schulz, Mohsen Mesgar, Krishnkant Swarnkar, Edwin Simpson, and Iryna Gurevych. 2019. Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systems. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 1634–1647, Minneapolis, Minnesota. Association for Computational Linguistics.
Cite (Informal):
Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systems (Eger et al., NAACL 2019)
Copy Citation:
PDF:
https://aclanthology.org/N19-1165.pdf
Presentation:
 N19-1165.Presentation.pptx
Video:
 https://vimeo.com/364693490