Pervasive Attention: 2D Convolutional Neural Networks for Sequence-to-Sequence Prediction

Maha Elbayad, Laurent Besacier, Jakob Verbeek


Abstract
Current state-of-the-art machine translation systems are based on encoder-decoder architectures, that first encode the input sequence, and then generate an output sequence based on the input encoding. Both are interfaced with an attention mechanism that recombines a fixed encoding of the source tokens based on the decoder state. We propose an alternative approach which instead relies on a single 2D convolutional neural network across both sequences. Each layer of our network re-codes source tokens on the basis of the output sequence produced so far. Attention-like properties are therefore pervasive throughout the network. Our model yields excellent results, outperforming state-of-the-art encoder-decoder systems, while being conceptually simpler and having fewer parameters.
Anthology ID:
K18-1010
Volume:
Proceedings of the 22nd Conference on Computational Natural Language Learning
Month:
October
Year:
2018
Address:
Brussels, Belgium
Editors:
Anna Korhonen, Ivan Titov
Venue:
CoNLL
SIG:
SIGNLL
Publisher:
Association for Computational Linguistics
Note:
Pages:
97–107
Language:
URL:
https://aclanthology.org/K18-1010
DOI:
10.18653/v1/K18-1010
Bibkey:
Cite (ACL):
Maha Elbayad, Laurent Besacier, and Jakob Verbeek. 2018. Pervasive Attention: 2D Convolutional Neural Networks for Sequence-to-Sequence Prediction. In Proceedings of the 22nd Conference on Computational Natural Language Learning, pages 97–107, Brussels, Belgium. Association for Computational Linguistics.
Cite (Informal):
Pervasive Attention: 2D Convolutional Neural Networks for Sequence-to-Sequence Prediction (Elbayad et al., CoNLL 2018)
Copy Citation:
PDF:
https://preview.aclanthology.org/improve-issue-templates/K18-1010.pdf
Code
 elbayadm/attn2d +  additional community code