Research map / Label-efficient and robust learning / Domain adaptation and generalization
MetaMT, a Meta Learning Method Leveraging Multiple Domain Data for Low Resource 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 |
|---|---|---|---|
| Use MetaMT instead of Transformer + Fine Tune | En-Es domain adaptation · EU Para, UN Data, GlobalVoices, JRCAcquis, EU Bookshops, OpenSub, Medline, and Electronic Health Record | BLEU-4: 37.10 -> 39.02; 54.81 -> 55.13; 45.24 -> 47.73; 60.07 -> 61.04; 22.29 -> 22.74; 27.31 -> 29.43; 57.58 -> 59.06; 40.61 -> | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · EU Para | BLEU-4: 38.32 -> 39.02 | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · UN Data | BLEU-4: 55.01 -> 55.13 | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · GlobalVoices | BLEU-4: 46.95 -> 47.73 | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · JRCAcquis | BLEU-4: 60.93 -> 61.04 | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · EU Bookshops | BLEU-4: 22.25 -> 22.74 | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · OpenSub | BLEU-4: 28.97 -> 29.43 | Helped |
| Add encoder-side embedding projection (ablation) | MetaMT · Medline | BLEU-4: 58.77 -> 59.06 | Helped |
| Add decoder-side embedding projection (ablation) | MetaMT · EU Para | BLEU-4: 37.34 -> 39.02 | Helped |
| Add decoder-side embedding projection (ablation) | MetaMT · UN Data | BLEU-4: 54.92 -> 55.13 | Helped |
| Add decoder-side embedding projection (ablation) | MetaMT · GlobalVoices | BLEU-4: 45.73 -> 47.73 | Helped |
| Add decoder-side embedding projection (ablation) | MetaMT · JRCAcquis | BLEU-4: 60.52 -> 61.04 | Helped |
Closest papers by meaning
- Meta-Curriculum Learning for Domain Adaptation in Neural Machine Translation (AAAI 2021 · similarity 0.93)
- Towards Reliable Neural Machine Translation with Consistency-Aware Meta-Learning (AAAI 2023 · similarity 0.91)
- Meta Back-Translation (ICLR 2021 · similarity 0.89)
- Joint Training for Neural Machine Translation Models with Monolingual Data (AAAI 2018 · similarity 0.89)
- From Fully Trained to Fully Random Embeddings: Improving Neural Machine Translation with Compact Word Embedding Tables (AAAI 2022 · similarity 0.89)
- A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models (ICLR 2024 · similarity 0.88)
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.