Review of Research Progress on Small Sample Learning

The neural network model has achieved very good results in the field of computer vision, but the requirements for the amount of data are directly proportional to performance. Data hungry is not only the cause of the high performance of the neural network model, but also a hindrance to its further development. There is not a lot of available data in real-world application scenarios, so how to obtain better performance and more robust models in a small sample environment becomes the goal of less sample learning research. This paper reviews the past small sample learning methods, sorts out its performance and theoretical principles, and points out the existing deficiencies and proposes possible future research directions.