 Transfer learning is a technique which uses knowledge gained from one task to improve another related task. In this paper, the authors use transfer learning to solve the problem of limited labeled SAR target data. They first train a convolutional autoencoder, SCAE, on a large set of unlabeled SAR scene images. This trained model is then used as a starting point for a new model which is trained on both the unlabeled SAR scene images and labeled SAR target data. The authors found that this approach improved the accuracy of the classifier compared to a baseline model without using any transferred knowledge. Additionally, they introduced a feedback loop which further improved the accuracy of the classifier. This article was authored by Zhong Linghuang, Zhong Xupan, and Bin Lei.