MetaMind Neural Machine Translation System for WMT 2016
By James Bradbury, Richard Socher. 2016
17 citations recorded by the source corpus.
Neural Machine Translation (NMT) systems, introduced only in 2013, have achieved state of the art results in many MT tasks. MetaMind’s submissions to WMT ’16 seek to push the state of the art in one such task, English → German news-domain translation. We integrate promising recent developments in NMT, including subword splitting and back-translation for monolingual data augmentation, and introduce the Y-LSTM, a novel neural translation architecture.
Subfields: Artificial Intelligence, Natural Language Processing, Model Architectures
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Yonatan Belinkov
- Richard Socher
- Graham Neubig
- Luke S. Zettlemoyer
- Mohit Bansal
- Percy Liang
- Caiming Xiong
- Kyunghyun Cho
In Natural Language Processing
- Yonatan Belinkov
- Graham Neubig
- Richard Socher
- Michael Auli
- Luke S. Zettlemoyer
- Caiming Xiong
- Kyunghyun Cho
- Mohit Bansal
In Model Architectures
- Yonatan Belinkov
- Richard Socher
- Michael Auli
- Caiming Xiong
- David Grangier
- Stephen Merity
- Yann Dauphin
- Hassan Sajjad
Citations in the corpus
- A Convolutional Encoder Model for Neural Machine Translation (2016, background): “…recurrent neural network (RNN) into a variable length representation and then generates the translation left-to-right with another RNN where both components interface via a soft-attention mechanism (Bahdanau et al., 2015; Luong et al., 2015a; Bradbury and Socher, 2016; Sennrich et al., 2016a).”
- Quasi-Recurrent Neural Networks (2016, methodology): “RNNs are also commonly the basic building block for more complex models for tasks such as machine translation (Bahdanau et al., 2015; Luong et al., 2015; Bradbury & Socher, 2016) or question answering (Kumar et al., 2016; Xiong et al., 2016).”
- One Size Does Not Fit All: Comparing NMT Representations of Different Granularities (2019, methodology): “A less popular solution is to use characters as the basic unit (Chung et al., 2016; Lee et al., 2017), and in the case of morphologically complex languages, yet another alternative is to reduce the vocabulary size by using unsupervised morpheme segmentation (Bradbury and Socher, 2016).”
- Linguistically Motivated Vocabulary Reduction for Neural Machine Translation from Turkish to English (2017, background): “Another study tried to overcome this limitation by us-ing the Baseline variant of Morfessor (Creutz and Lagus, 2005b), which allows to reach a vocabulary size set prior to segmentation (Bradbury and Socher, 2016).”