Dual Contrastive Learning: Text Classification via Label-Aware Data Augmentation

TL;DR

This paper introduces DualCL, a dual contrastive learning framework that enhances text classification by simultaneously learning sample features and classifier parameters through label-aware data augmentation, improving accuracy especially in low-resource scenarios.

Contribution

The paper proposes a novel dual contrastive learning approach that jointly learns input features and classifier parameters, addressing the challenge of adapting contrastive learning to supervised text classification.

Findings

Improves classification accuracy on five benchmark datasets.

Effectively learns discriminative representations in low-resource settings.

Demonstrates the effectiveness of label-aware data augmentation.

Abstract

Contrastive learning has achieved remarkable success in representation learning via self-supervision in unsupervised settings. However, effectively adapting contrastive learning to supervised learning tasks remains as a challenge in practice. In this work, we introduce a dual contrastive learning (DualCL) framework that simultaneously learns the features of input samples and the parameters of classifiers in the same space. Specifically, DualCL regards the parameters of the classifiers as augmented samples associating to different labels and then exploits the contrastive learning between the input samples and the augmented samples. Empirical studies on five benchmark text classification datasets and their low-resource version demonstrate the improvement in classification accuracy and confirm the capability of learning discriminative representations of DualCL.