SSMix: Saliency-Based Span Mixup for Text Classification
This work addresses the problem of data augmentation for text classification, offering a novel approach that improves performance over existing methods, though it is incremental in the context of mixup techniques.
The authors tackled the challenge of applying mixup data augmentation to NLP tasks by proposing SSMix, a saliency-based span mixup method that operates on input text, and demonstrated that it outperforms hidden-level mixup methods on various text classification benchmarks.
Data augmentation with mixup has shown to be effective on various computer vision tasks. Despite its great success, there has been a hurdle to apply mixup to NLP tasks since text consists of discrete tokens with variable length. In this work, we propose SSMix, a novel mixup method where the operation is performed on input text rather than on hidden vectors like previous approaches. SSMix synthesizes a sentence while preserving the locality of two original texts by span-based mixing and keeping more tokens related to the prediction relying on saliency information. With extensive experiments, we empirically validate that our method outperforms hidden-level mixup methods on a wide range of text classification benchmarks, including textual entailment, sentiment classification, and question-type classification. Our code is available at https://github.com/clovaai/ssmix.