Research map / Label-efficient and robust learning / Recognition and long-tail
Multilingual Neural Machine Translation With Soft Decoupled Encoding
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 |
|---|---|---|---|
| replace Sub-sep Lookup with SDE word-level encoding | multilingual NMT · aze→English, bel→English, glg→English, slk→English | BLEU: 10.90 -> 11.82; 16.17 -> 18.71; 28.10 -> 30.30; 28.50 -> 28.77 | Helped |
| add the latent semantic embedding (ablation) | SDE · aze→English, bel→English, glg→English, slk→English | BLEU: 7.77 -> 11.82; 15.66 -> 18.71; 29.25 -> 30.30; 28.15 -> 28.77 | Helped |
| add the character n-gram lexical embedding (ablation) | SDE · aze→English, bel→English, glg→English, slk→English | BLEU: 4.57 -> 11.82; 8.03 -> 18.71; 13.77 -> 30.30; 7.08 -> 28.77 | Helped |
| add the language-specific transform (ablation) | SDE · aze→English | BLEU: 12.89 -> 11.82 | Hurt |
| add the language-specific transform (ablation) | SDE · slk→English | BLEU: 29.16 -> 28.77 | Hurt |
| add the language-specific transform (ablation) | SDE · bel→English | BLEU: 18.13 -> 18.71 | Helped |
| add the language-specific transform (ablation) | SDE · glg→English | BLEU: 30.07 -> 30.30 | No clear effect |
| use Sub-sep pieces as lexical units instead of words (ablation) | SDE · aze→English | BLEU: 11.82 -> 12.37 | Helped |
| use Sub-sep pieces as lexical units instead of words (ablation) | SDE · bel→English, glg→English, slk→English | BLEU: 18.71 -> 16.29; 30.30 -> 28.94; 28.77 -> 28.35 | Hurt |
| replace character n-grams with subword pieces (ablation) | SDE-sub · aze→English, slk→English | BLEU: 11.82 -> 12.03; 28.77 -> 28.86 | No clear effect |
| replace character n-grams with subword pieces (ablation) | SDE-sub · bel→English | BLEU: 18.71 -> 18.16 | Hurt |
| replace character n-grams with subword pieces (ablation) | SDE-sub · glg→English | BLEU: 30.30 -> 31.16 | Helped |
Closest papers by meaning
- Evaluating the Cross-Lingual Effectiveness of Massively Multilingual Neural Machine Translation (AAAI 2020 · similarity 0.89)
- From Fully Trained to Fully Random Embeddings: Improving Neural Machine Translation with Compact Word Embedding Tables (AAAI 2022 · similarity 0.89)
- Unsupervised Neural Machine Translation (ICLR 2018 · similarity 0.88)
- Neural Machine Translation with Soft Prototype (NeurIPS 2019 · similarity 0.87)
- Neural Machine Translation with Byte-Level Subwords (AAAI 2020 · similarity 0.87)
- Multi-Channel Encoder for Neural Machine Translation (AAAI 2018 · similarity 0.87)
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.