Adversarial Word Dilution as Text Data Augmentation in Low-Resource Regime

TL;DR

This paper introduces Adversarial Word Dilution (AWD), a novel data augmentation technique for low-resource text classification that creates hard positive examples by adversarially diluting strong positive words, leading to improved model performance.

Contribution

The paper proposes AWD, a new adversarial augmentation method that generates interpretable hard positive examples by diluting key words, outperforming existing augmentation techniques in low-resource settings.

Findings

AWD outperforms state-of-the-art augmentation methods on benchmark datasets.

Generated augmentations are interpretable and adaptable to new data.

AWD enhances low-resource text classification accuracy.

Abstract

Data augmentation is widely used in text classification, especially in the low-resource regime where a few examples for each class are available during training. Despite the success, generating data augmentations as hard positive examples that may increase their effectiveness is under-explored. This paper proposes an Adversarial Word Dilution (AWD) method that can generate hard positive examples as text data augmentations to train the low-resource text classification model efficiently. Our idea of augmenting the text data is to dilute the embedding of strong positive words by weighted mixing with unknown-word embedding, making the augmented inputs hard to be recognized as positive by the classification model. We adversarially learn the dilution weights through a constrained min-max optimization process with the guidance of the labels. Empirical studies on three benchmark datasets show that AWD can generate more effective data augmentations and outperform the state-of-the-art text data augmentation methods. The additional analysis demonstrates that the data augmentations generated by AWD are interpretable and can flexibly extend to new examples without further training.