Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Neural Dialect Translation with Commonality and Diversity Modeling
Yu Wan, Baosong Yang, Derek F. Wong, Lidia S. Chao, Haihua Du, Ben C.H. Ao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 layer coordination | Unsupervised NMT · Can⇒Man | BLEU: 42.90 -> 48.45 | Improved |
| Add layer coordination | Unsupervised NMT · Man⇒Can | BLEU: 42.39 -> 43.11 | Improved |
| Add pivot-private embedding | Unsupervised NMT · Can⇒Man | BLEU: 42.90 -> 52.74 | Improved |
| Add pivot-private embedding | Unsupervised NMT · Man⇒Can | BLEU: 42.39 -> 46.69 | Improved |
| Add pivot-private embedding and layer coordination | Unsupervised NMT · Can⇒Man | BLEU: 42.90 -> 54.95 | Improved |
| Add pivot-private embedding and layer coordination | Unsupervised NMT · Man⇒Can | BLEU: 42.39 -> 47.45 | Improved |
| Use an intermediate pivot-embedding dimension rather than no shared dimensions or all shared dimensions (ablation) | Pivot-Private Embedding · Can⇒Man and Man⇒Can dev | BLEU | Improved |
| Set pivot embedding dimensionality to 256 rather than 128 (ablation) | Pivot-Private Embedding · Can⇒Man and Man⇒Can dev | BLEU | No clear effect |
| Increase the number of shared layers (ablation) | Layer Coordination · Can⇒Man and Man⇒Can dev | BLEU | Improved |
| Share all encoder and decoder layers (ablation) | Layer Coordination · Can⇒Man and Man⇒Can dev | BLEU | Improved |
| Add layer coordination (ablation) | Unsupervised dialect translation · Training convergence | Training stability | Improved |
Closest papers by meaning
- Unsupervised Neural Machine Translation (ICLR 2018 · similarity 0.89)
- Unsupervised Machine Translation Using Monolingual Corpora Only (ICLR 2018 · similarity 0.88)
- Cross-lingual Language Model Pretraining (NeurIPS 2019 · similarity 0.87)
- Joint Training for Neural Machine Translation Models with Monolingual Data (AAAI 2018 · similarity 0.87)
- Cross-model Back-translated Distillation for Unsupervised Machine Translation (ICML 2021 · similarity 0.87)
- A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models (ICLR 2024 · similarity 0.85)
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.
