In order to improve the performance of radar target recognition (RTR) system, machine learning is applied to the field of target recognition. This paper summarizes the advantages and disadvantages of four mainstream methods including deep learning, neural network, statistical learning and ensemble learning, analyzes the optimization measures, and proposes that, as a compound process, can integrate the advantages of machine learning methods in target recognition. The results show that the organic combination of various methods in machine learning is the development trend of based on machine learning in the future.