Research map / Label-efficient and robust learning / Recognition and long-tail
Meta Back-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 |
|---|---|---|---|
| Adapt the backward model throughout forward-model training using meta-validation performance | Sampling-based back-translation · WMT En-De 2014 | BLEU: +1.66 BLEU | Helped |
| Adapt the backward model throughout forward-model training using meta-validation performance | Sampling-based back-translation · WMT En-Fr 2014 | BLEU: +0.51 BLEU | Helped |
| Adapt the backward model using meta-validation performance when initialized with related-language parallel data | Multilingual low-resource translation · az-en, be-en, gl-en, sk-en | BLEU: up to +1.20 BLEU | Helped |
| Optimize the backward model with reinforcement learning using language-model and reconstruction rewards instead of keeping an MLE model fixed | DualNMT · Multilingual and WMT translation tasks | BLEU | Hurt |
| Train on monolingual data with sampling-based back-translation rather than parallel data only | MLE back-translation · WMT En-De 2014 | BLEU | Helped |
| Train using UNMT rather than MetaBT | WMT En-De 2014 · WMT En-De 2014 | BLEU: 28.76 vs 28.73 | Hurt |
Closest papers by meaning
- MetaMT, a Meta Learning Method Leveraging Multiple Domain Data for Low Resource Machine Translation (AAAI 2020 · similarity 0.89)
- Towards Reliable Neural Machine Translation with Consistency-Aware Meta-Learning (AAAI 2023 · similarity 0.89)
- Data Diversification: A Simple Strategy For Neural Machine Translation (NeurIPS 2020 · similarity 0.88)
- Unsupervised Neural Machine Translation with SMT as Posterior Regularization (AAAI 2019 · similarity 0.88)
- Unsupervised Neural Machine Translation (ICLR 2018 · similarity 0.88)
- A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models (ICLR 2024 · similarity 0.86)
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.