Multi-Source Neural Translation
By Barret Zoph, Kevin Knight. 2016
334 citations recorded by the source corpus.
We build a multi-source machine translation model and train it to maximize the probability of a target English string given French and German sources. Using the neural encoder-decoder framework, we explore several combination methods and report up to +4.8 Bleu increases on top of a very strong attention-based neural translation model.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Machine Translation
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
- Graham Neubig
- Felix Stahlberg
- Kyunghyun Cho
- Mohit Bansal
- Luke S. Zettlemoyer
- Jianfeng Gao
- Yonatan Belinkov
- Percy Liang
In Computer Vision and Pattern Recognition
- Felix Stahlberg
- Graham Neubig
- Luc Van Gool
- Dacheng Tao
- Kyunghyun Cho
- Trevor Darrell
- Xiaogang Wang
- Andrew Zisserman
In Machine Translation
- Felix Stahlberg
- Raj Dabre
- Chenhui Chu
- Markus Freitag
- Jiatao Gu
- Graham Neubig
- Matteo Negri
- Marco Turchi
Citations in the corpus
- Complete Multilingual Neural Machine Translation (2020, neutral): “Another approach is multi-source NMT (Zoph and Knight, 2016).”
- Zero-Resource Translation with Multi-Lingual Neural Machine Translation (2016, methodology): “One may decide to introduce a mechanism for exploiting these (Zoph and Knight, 2016), or as we present here, it may not be necessary at all to do so.”
- Toward Multilingual Neural Machine Translation with Universal Encoder and Decoder (2016, methodology): “Our encoder then acts similarly to the one of multi-source approach[12], collecting additional information from other sources for better translations, but with a much simpler embedding function.”
- Zero-Shot Cross-lingual Classification Using Multilingual Neural Machine Translation (2018, methodology): “Multilingual SNLI We extended the proposed Encoder-Classifier model to a multi-source model (Zoph and Knight 2016) since SNLI is an inference task of relations between two input sentences, “premise” and “hypothesis”.”
- Marian: Fast Neural Machine Translation in C++ (2018, methodology): “These models are based on multi-source neural translation models introduced by Zoph and Knight (2016).”
- Paraphrasing Revisited with Neural Machine Translation (2017, methodology): “Inspired by multi-way machine translation which has shown performance gains over single-pair models (Zoph and Knight, 2016; Dong et al., 2015; Firat et al., 2016a), we also explore an alternative pivoting technique which uses multiple languages rather than a single one.”
- Exploiting Source-side Monolingual Data in Neural Machine Translation (2016, background): “Zoph et al. (2016), Zoph and Knight (2016) and Firat et al. (2016) further deal with more complicated cases (e.g. multi-source languages).”
- A neural interlingua for multilingual machine translation (2018, background): “The many-to-one approach explored in Zoph and Knight (2016) primarily considers the trilingual case, where a multi-parallel corpus is available, and uses 2 encoders simultaneously to provide the source context for the decoder.”
- A Survey of Cross-lingual Word Embedding Models (2017, neutral): “…multi-source learning and multi-source transfers: this results in a more general model and is less prone to data sparseness (McDonald et al., 2011; Agić, Johannsen, Plank, Alonso, Schluter, & Søgaard, 2016; Guo, Che, Yarowsky, Wang, & Liu, 2016; Zoph & Knight, 2016; Firat, Cho, & Bengio, 2016).”
- Evaluating Discourse Phenomena in Neural Machine Translation (2017, methodology): “We have presented an evaluation of discourse-level NMT models through the use of two discourse test sets targeted at coreference and lexical coherence/cohesion.”
- Two are Better than One: An Ensemble of Retrieval- and Generation-Based Dialog Systems (2016, methodology): “The query, along with the candidate reply, is then fed to an utterance generator based on the “bi-sequence to sequence” ( biseq2seq ) model [30].”
- Listen and Translate: A Proof of Concept for End-to-End Speech-to-Text Translation (2016, neutral): “4 Our implementation also allows pre-training, multiple encoders (Zoph and Knight, 2016) and multi-task training (Luong et al., 2016).”
- Contextual Parameter Generation for Universal Neural Machine Translation (2018, background): “Zoph and Knight (2016) flipped this idea with a many-to-one translation model, however requiring the presence of a multi-way parallel corpus between all the languages, which is difficult to obtain.”
- Emerging Language Spaces Learned From Massively Multilingual Corpora (2018, background): “Zoph and Knight [21] propose a multi-source model and [22] introduces a character-level encoder that is shared across several source languages.”
- An Empirical Analysis of NMT-Derived Interlingual Embeddings and Their Use in Parallel Sentence Identification (2017, background): “They use multiple encoders and/or decoders with multiple or shared attention mechanisms [10], [11], [12], [13], [14].”
- Learning Joint Multilingual Sentence Representations with Neural Machine Translation (2017, background): “NMT has been also extended to handle several source and/or target languages at once, with the goal of achieving better translation quality than with separately trained NMT systems, in particular for under resourced languages, see for instance (Dong et al., 2015; Zoph and Knight, 2016; Luong et al., ”
- Search Engine Guided Non-Parametric Neural Machine Translation (2017, methodology): “The proposed TM-NMT has been largely motivated by recently proposed multilingual attention-based neural machine translation models (see, e.g., Firat et al., 2016a; Zoph and Knight, 2016).”