Using error decay prediction to overcome practical issues of deep active learning for named entity recognition.
Haw-Shiuan ChangShankar VembuSunil MohanRheeya UppaalAndrew McCallumPublished in: Mach. Learn. (2020)
Keyphrases
- practical issues
- named entity recognition
- active learning
- semi supervised
- information extraction
- named entities
- natural language processing
- maximum entropy
- hand coded
- prediction accuracy
- text summarization
- conditional random fields
- annotated corpus
- relation extraction
- pairwise
- classifier ensemble
- machine learning
- sequence labeling
- maximum entropy classifier
- named entity disambiguation
- learning algorithm
- question answering
- labeled data
- learning process
- natural language
- unsupervised learning
- supervised learning
- object recognition