LexicalAT: Lexical-Based Adversarial Reinforcement Training for Robust Sentiment Classification

Jingjing Xu, Liang Zhao, Hanqi Yan, Qi Zeng, Yun Liang, Xu Sun


Abstract
Recent work has shown that current text classification models are fragile and sensitive to simple perturbations. In this work, we propose a novel adversarial training approach, LexicalAT, to improve the robustness of current classification models. The proposed approach consists of a generator and a classifier. The generator learns to generate examples to attack the classifier while the classifier learns to defend these attacks. Considering the diversity of attacks, the generator uses a large-scale lexical knowledge base, WordNet, to generate attacking examples by replacing some words in training examples with their synonyms (e.g., sad and unhappy), neighbor words (e.g., fox and wolf), or super-superior words (e.g., chair and armchair). Due to the discrete generation step in the generator, we use policy gradient, a reinforcement learning approach, to train the two modules. Experiments show LexicalAT outperforms strong baselines and reduces test errors on various neural networks, including CNN, RNN, and BERT.
Anthology ID:
D19-1554
Volume:
Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)
Month:
November
Year:
2019
Address:
Hong Kong, China
Editors:
Kentaro Inui, Jing Jiang, Vincent Ng, Xiaojun Wan
Venues:
EMNLP | IJCNLP
SIG:
SIGDAT
Publisher:
Association for Computational Linguistics
Note:
Pages:
5518–5527
Language:
URL:
https://aclanthology.org/D19-1554
DOI:
10.18653/v1/D19-1554
Bibkey:
Cite (ACL):
Jingjing Xu, Liang Zhao, Hanqi Yan, Qi Zeng, Yun Liang, and Xu Sun. 2019. LexicalAT: Lexical-Based Adversarial Reinforcement Training for Robust Sentiment Classification. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 5518–5527, Hong Kong, China. Association for Computational Linguistics.
Cite (Informal):
LexicalAT: Lexical-Based Adversarial Reinforcement Training for Robust Sentiment Classification (Xu et al., EMNLP-IJCNLP 2019)
Copy Citation:
PDF:
https://aclanthology.org/D19-1554.pdf