Порівняння методів
Переглядайте обрані методи поруч; рядки з відмінностями підсвічено.
| Адаптивне розпізнавання іменованих сутностей× | Класифікація на основі BERT з адаптацією до предметної області× | |
|---|---|---|
| Галузь | Глибоке навчання | Глибоке навчання |
| Родина | Machine learning | Machine learning |
| Рік появи≠ | 2006–2020 | 2019–2020 |
| Автор методу≠ | Multiple contributors (Blitzer et al., 2006; Daumé, 2007; Lee et al., 2020) | Gururangan et al. (2020); earlier domain-specific instances include Lee et al. (2020) — BioBERT |
| Тип≠ | Sequence labeling with domain adaptation | Domain-adaptive pre-training followed by supervised fine-tuning |
| Основоположне джерело≠ | Lee, J., Yoon, W., Kim, S., Kim, D., Kim, S., So, C. H., & Kang, J. (2020). BioBERT: a pre-trained biomedical language representation model for biomedical text mining. Bioinformatics, 36(4), 1234–1240. DOI ↗ | Gururangan, S., Marasovic, A., Swayamdipta, S., Lo, K., Beltagy, I., Downey, D., & Smith, N. A. (2020). Don't Stop Pretraining: Adapt Language Models to Domains and Tasks. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL 2020), 8342–8360. DOI ↗ |
| Інші назви | DA-NER, cross-domain NER, domain-adaptive NER, domain-transfer named entity recognition | DAPT BERT classification, domain-adaptive pre-training, domain-specific BERT fine-tuning, BERT DAPT |
| Пов'язані≠ | 5 | 6 |
| Підсумок≠ | Domain-adaptive Named Entity Recognition (DA-NER) applies named entity recognition to a target domain by transferring or adapting a model trained on a source domain, using techniques such as domain-specific pre-training, adversarial alignment, or feature augmentation. It addresses the performance collapse that standard NER models suffer when deployed outside their training domain. | Domain-adaptive BERT-based classification extends the standard fine-tuning pipeline by first continuing BERT's masked-language-model pre-training on a large corpus of in-domain unlabeled text, then fine-tuning the adapted model on labeled examples for the target classification task. This two-stage approach closes the vocabulary and distributional gap between BERT's general pre-training corpus and specialized domains such as biomedicine, law, finance, or social-media text. |
| ScholarGateНабір даних ↗ |
|
|