A team at ORNL has demonstrated that the combination of transfer learning and semi-supervised learning can significantly reduce the amount of labeled data required to obtain strong performance in biomedical named entity recognition (NER) tasks.
Menu