SwitchOut: an Efficient Data Augmentation Algorithm for Neural Machine Translation
This addresses the need for efficient data augmentation in neural machine translation, offering a simple yet effective solution that is incremental over existing methods like word dropout.
The paper tackles the problem of data augmentation for neural machine translation by proposing SwitchOut, a method that randomly replaces words in source and target sentences with other random words from their vocabularies, resulting in consistent improvements of about 0.5 BLEU across three translation datasets.
In this work, we examine methods for data augmentation for text-based tasks such as neural machine translation (NMT). We formulate the design of a data augmentation policy with desirable properties as an optimization problem, and derive a generic analytic solution. This solution not only subsumes some existing augmentation schemes, but also leads to an extremely simple data augmentation strategy for NMT: randomly replacing words in both the source sentence and the target sentence with other random words from their corresponding vocabularies. We name this method SwitchOut. Experiments on three translation datasets of different scales show that SwitchOut yields consistent improvements of about 0.5 BLEU, achieving better or comparable performances to strong alternatives such as word dropout (Sennrich et al., 2016a). Code to implement this method is included in the appendix.