Attention Meets Perturbations: Robust and Interpretable Attention with Adversarial Training

TL;DR

This paper introduces adversarial training techniques for attention mechanisms in NLP, significantly improving robustness and interpretability across multiple datasets.

Contribution

It proposes Attention AT and Attention iAT, novel adversarial training methods that enhance attention robustness and interpretability in NLP models.

Findings

Attention iAT outperformed in 9 of 10 tasks

Attention attention correlated more with word importance

Techniques are less sensitive to perturbation size

Abstract

Although attention mechanisms have been applied to a variety of deep learning models and have been shown to improve the prediction performance, it has been reported to be vulnerable to perturbations to the mechanism. To overcome the vulnerability to perturbations in the mechanism, we are inspired by adversarial training (AT), which is a powerful regularization technique for enhancing the robustness of the models. In this paper, we propose a general training technique for natural language processing tasks, including AT for attention (Attention AT) and more interpretable AT for attention (Attention iAT). The proposed techniques improved the prediction performance and the model interpretability by exploiting the mechanisms with AT. In particular, Attention iAT boosts those advantages by introducing adversarial perturbation, which enhances the difference in the attention of the sentences. Evaluation experiments with ten open datasets revealed that AT for attention mechanisms, especially Attention iAT, demonstrated (1) the best performance in nine out of ten tasks and (2) more interpretable attention (i.e., the resulting attention correlated more strongly with gradient-based word importance) for all tasks. Additionally, the proposed techniques are (3) much less dependent on perturbation size in AT. Our code is available at https://github.com/shunk031/attention-meets-perturbation