Research on Transfer Learning Technology in Natural Language Processing

摘要

In the field of natural language processing, traditional machine learning methods rely on a large amount of labeled data to obtain better results and require training data and test data to satisfy the assumption of independent and identically distributed. Application of transfer learning technology provides a good solution to the problems above and promotes the research progress of natural language processing. This paper studied the application of transfer learning in natural language processing, made the definition of transfer learning, classified the transfer learning in natural language processing, and prospected the development of transfer learning in natural language processing.