Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Neural Machine Translation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add backtranslation to denoising training (ablation) | Unsupervised NMT · WMT14 newstest2014 FR-EN | BLEU: 7.28 -> 15.56 | Helped |
| add backtranslation to denoising training (ablation) | Unsupervised NMT · WMT14 newstest2014 EN-FR | BLEU: 5.33 -> 15.13 | Helped |
| add backtranslation to denoising training (ablation) | Unsupervised NMT · WMT14 newstest2014 DE-EN | BLEU: 3.64 -> 10.21 | Helped |
| add backtranslation to denoising training (ablation) | Unsupervised NMT · WMT14 newstest2014 EN-DE | BLEU: 2.40 -> 6.55 | Helped |
| add denoising without backtranslation (ablation) | Unsupervised NMT · WMT14 newstest2014 FR-EN | BLEU: 9.98 -> 7.28 | Hurt |
| add denoising without backtranslation (ablation) | Unsupervised NMT · WMT14 newstest2014 EN-FR | BLEU: 6.25 -> 5.33 | Hurt |
| use BPE instead of word-level tokens (ablation) | Unsupervised NMT with backtranslation · WMT14 newstest2014 FR-EN | BLEU: 15.56 -> 15.56 | No clear effect |
| use BPE instead of word-level tokens (ablation) | Unsupervised NMT with backtranslation · WMT14 newstest2014 EN-FR | BLEU: 15.13 -> 14.36 | Hurt |
| use BPE instead of word-level tokens (ablation) | Unsupervised NMT with backtranslation · WMT14 newstest2014 DE-EN | BLEU: 10.21 -> 10.16 | No clear effect |
| use BPE instead of word-level tokens (ablation) | Unsupervised NMT with backtranslation · WMT14 newstest2014 EN-DE | BLEU: 6.55 -> 6.89 | Helped |
| add 10,000 parallel sentence pairs | Proposed unsupervised NMT · WMT14 newstest2014 FR-EN | BLEU: 15.56 -> 18.57 | Helped |
| add 100,000 parallel sentence pairs | Proposed unsupervised NMT · WMT14 newstest2014 FR-EN | BLEU: 15.56 -> 21.81 | Helped |
Closest papers by meaning
- Unsupervised Machine Translation Using Monolingual Corpora Only (ICLR 2018 · similarity 0.94)
- From Fully Trained to Fully Random Embeddings: Improving Neural Machine Translation with Compact Word Embedding Tables (AAAI 2022 · similarity 0.91)
- Unsupervised Neural Machine Translation with SMT as Posterior Regularization (AAAI 2019 · similarity 0.91)
- Contrastive Clustering to Mine Pseudo Parallel Data for Unsupervised Translation (ICLR 2022 · similarity 0.90)
- Empirical Regularization for Synthetic Sentence Pairs in Unsupervised Neural Machine Translation (AAAI 2021 · similarity 0.90)
- Joint Training for Neural Machine Translation Models with Monolingual Data (AAAI 2018 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.