2019

On Compositionality in Neural Machine Translation

Raunak, Vikas, Kumar, Vaibhav, Metze, Florian

Understand

We investigate two specific manifestations of compositionality in Neural Machine Translation (NMT) : (1) Productivity - the ability of the model to extend its predictions beyond the observed length in training data and (2) Systematicity - the ability of the model to systematically recombine known parts and rules.

  • We evaluate a standard Sequence to Sequence model on tests designed to assess these two properties in NMT.
  • We quantitatively demonstrate that inadequate temporal processing, in the form of poor encoder representations is a bottleneck for both Productivity and Systematicity.
  • We propose a simple pre-training mechanism which alleviates model performance on the two properties and leads to a significant improvement in BLEU scores.

Built on

  • Report on the 11th iwslt evaluation campaign, iwslt 2014

    M. Cettolo, J. Niehues, S. Stüker, L. Bentivogli, and M. Federico · 2014

    Earlier work this paper cites.

  • On the properties of neural machine translation: Encoder-decoder approaches

    Original

    K. Cho, B. Van Merriënboer, D. Bahdanau, and Y. Bengio · 2014

    Earlier work this paper cites.

  • Sequence to sequence learning with neural networks

    I. Sutskever, O. Vinyals, and Q. V. Le · 2014

    Earlier work this paper cites.

  • Neural machine translation by jointly learning to align and translate

    D. Bahdanau, K. Cho, and Y. Bengio · 2015

    Earlier work this paper cites.

  • Deep visual-semantic alignments for generating image descriptions

    A. Karpathy and L. Fei-Fei · 2015

    Earlier work this paper cites.

Similar

Then

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…