Regularization techniques for fine-tuning in neural machine translation
This addresses the problem of domain adaptation in neural machine translation for researchers and practitioners, but it is incremental as it builds on existing regularization methods.
The paper tackles overfitting in supervised domain adaptation for neural machine translation by investigating regularization techniques, including a novel method called tuneout, and finds improvements on IWSLT datasets for English->German and English->Russian, with a logarithmic relationship between in-domain training data and BLEU score gain.
We investigate techniques for supervised domain adaptation for neural machine translation where an existing model trained on a large out-of-domain dataset is adapted to a small in-domain dataset. In this scenario, overfitting is a major challenge. We investigate a number of techniques to reduce overfitting and improve transfer learning, including regularization techniques such as dropout and L2-regularization towards an out-of-domain prior. In addition, we introduce tuneout, a novel regularization technique inspired by dropout. We apply these techniques, alone and in combination, to neural machine translation, obtaining improvements on IWSLT datasets for English->German and English->Russian. We also investigate the amounts of in-domain training data needed for domain adaptation in NMT, and find a logarithmic relationship between the amount of training data and gain in BLEU score.