Research map / Architectures and efficiency / RNNs and normalization
Transfer Learning for Sequence Tagging with Hierarchical Recurrent Networks
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 cross-domain T-A transfer from PTB | Twitter POS tagging · Twitter POS/0.1 | POS accuracy: 74.80 -> 83.65 | Helped |
| add cross-domain T-A transfer from CoNLL 2003 | Twitter NER · Twitter NER/0.1 | NER F1: 34.65 -> 43.24 | Helped |
| add cross-application transfer from PTB | CoNLL 2003 NER · CoNLL 2003 NER/0.01 | NER F1: 68.64 -> 74.92 | Helped |
| add cross-application transfer from PTB | CoNLL 2000 chunking · CoNLL 2000/0.01 | chunk-level F1: 83.49 -> 86.73 | Helped |
| add cross-lingual transfer from Spanish | CoNLL 2003 NER · CoNLL 2003 NER/0.01 | NER F1: 68.64 -> 72.61 | Helped |
| add cross-lingual transfer from CoNLL 2003 | Spanish NER · Spanish NER/0.01 | NER F1: 59.84 -> 60.43 | Helped |
| add cross-domain transfer from PTB | Genia POS tagging · Genia/0.001 | POS accuracy: 83.26 -> 92.62 | Helped |
| use T-B instead of T-A, sharing fewer parameters (ablation) | transfer from PTB to Genia POS tagging · Genia/0.001 | POS accuracy: 92.62 -> 89.77 | Hurt |
| use T-C instead of T-A, sharing fewer parameters (ablation) | transfer from PTB to Genia POS tagging · Genia/0.001 | POS accuracy: 92.62 -> 84.65 | Hurt |
| add transfer between sequence-tagging tasks | CoNLL 2000 chunking · CoNLL 2000 | chunk-level F1: 94.66 -> 95.41 | Helped |
| add transfer between sequence-tagging tasks | CoNLL 2003 NER · CoNLL 2003 | NER F1: 91.20 -> 91.26 | Helped |
| add transfer between sequence-tagging tasks | Spanish NER · Spanish NER | NER F1: 84.69 -> 85.77 | Helped |
Closest papers by meaning
- Transfer Learning for Sequences via Learning to Collocate (ICLR 2019 · similarity 0.89)
- Transfer Learning for Sequence Labeling Using Source Model and Target Data (AAAI 2019 · similarity 0.89)
- Weakly Supervised Sequence Tagging from Noisy Rules (AAAI 2020 · similarity 0.89)
- Empower Sequence Labeling with Task-Aware Neural Language Model (AAAI 2018 · similarity 0.89)
- Sub-Task Decomposition Enables Learning in Sequence to Sequence Tasks (ICLR 2023 · similarity 0.87)
- PoNet: Pooling Network for Efficient Token Mixing in Long Sequences (ICLR 2022 · 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.