Text Augmentation in a Multi-Task View

TL;DR

This paper introduces a multi-task view of data augmentation for text classification, training on original and augmented data simultaneously, which enhances robustness and performance over traditional augmentation methods.

Contribution

It proposes a novel multi-task framework for data augmentation that allows stronger augmentation and improves model robustness and accuracy.

Findings

MTV outperforms traditional augmentation in experiments.

Stronger augmentation levels are feasible with MTV.

Robust performance gains across multiple datasets.

Abstract

Traditional data augmentation aims to increase the coverage of the input distribution by generating augmented examples that strongly resemble original samples in an online fashion where augmented examples dominate training. In this paper, we propose an alternative perspective -- a multi-task view (MTV) of data augmentation -- in which the primary task trains on original examples and the auxiliary task trains on augmented examples. In MTV data augmentation, both original and augmented samples are weighted substantively during training, relaxing the constraint that augmented examples must resemble original data and thereby allowing us to apply stronger levels of augmentation. In empirical experiments using four common data augmentation techniques on three benchmark text classification datasets, we find that the MTV leads to higher and more robust performance improvements than traditional augmentation.