As a new classification model, deep learning has been paid more and more attention by researchers in recent years and has been successfully applied in many fields. In the fields of bioinformatics and robotics, it is very difficult to construct large-scale well-annotated data sets because of the high cost of data collection and annotation, which limits the development of data sets. Migration learning doesn't require that the training data must be independent and distributed with the test data, which inspires us to use migration learning to solve the problem of insufficient training data. This paper summarizes the research status and application of deep neural network in transfer learning. We define deep transfer learning and its classification, and review the research work based on deep transfer learning technology in recent years.