Research map / Detection and segmentation / CNN features
Switch-LSTMs for Multi-Criteria Chinese Word Segmentation
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 Switch-LSTMs instead of the multi-task framework | Multi-criteria CWS · Eight CWS datasets (average) | F: 94.86 -> 96.12 | Helped |
| use multi-criteria training instead of single-criterion training | Switch-LSTMs · Eight CWS datasets (average) | F: 94.76 -> 96.12 | Helped |
| use more than 1 switch (ablation) | Switch-LSTMs · Eight CWS datasets (average) | F: 92.27 -> 95.53 | Helped |
| use 4 switches rather than 2 (ablation) | Switch-LSTMs · Eight CWS datasets (average) | F: 95.53 -> 95.84 | Helped |
| add task embeddings to switch inputs (ablation) | Switch-LSTMs · Eight CWS datasets (average) | F: 92.51 -> 95.84 | Helped |
| add switch state inputs (ablation) | Switch-LSTMs without task embedding · Eight CWS datasets (average) | F: 91.70 -> 92.51 | Helped |
| add input embeddings to switch inputs (ablation) | Switch-LSTMs · Eight CWS datasets (average) | F: 95.54 -> 95.84 | Helped |
| randomize switch selection during training (ablation) | Switch-LSTMs · Eight CWS datasets (average) | F: 95.84 -> 92.53 | Hurt |
| randomize switch selection during testing (ablation) | Switch-LSTMs · Eight CWS datasets (average) | F: 95.84 -> 91.29 | Hurt |
| transfer parameters from seven datasets and train only a new task embedding | Switch-LSTMs · New criterion, 1000 training instances (average) | F: 76.30 -> 92.76 | Helped |
Closest papers by meaning
- Neural Networks Incorporating Dictionaries for Chinese Word Segmentation (AAAI 2018 · similarity 0.87)
- Unsupervised Learning Helps Supervised Neural Word Segmentation (AAAI 2019 · similarity 0.87)
- BTS: A Bi-Lingual Benchmark for Text Segmentation in the Wild (CVPR 2022 · similarity 0.84)
- Morphological Segmentation with Window LSTM Neural Networks (AAAI 2016 · similarity 0.84)
- Vietnamese Words Are Not Constructed from Syllables: Rethinking the Role of Word Segmentation in Natural Language Processing for Vietnamese Texts (AAAI 2025 · similarity 0.81)
- SENCR: A Span Enhanced Two-Stage Network with Counterfactual Rethinking for Chinese NER (AAAI 2024 · similarity 0.81)
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.