Abstract: In materials genetic engineering, data-driven machine learning based techniques have drawn much attention as a powerful new tool in the field of magnesium alloys. The traditional empirical trial-and-error methods and the methods based on density functional theory have been difficult to meet the continuous development of materials science needs due to their high time cost and low efficiency. Based on statistics instead of solving physical equations, as long as the connection between descriptors and target properties is identified, machine learning can almost cut out any complicated experimental process and quickly predict material properties at a low cost. In materials science, magnesium and its alloys have tremendous potential in aerospace, automotive and other fields due to their low density and high specific strength. However, due to the problems of the different alloying elements effect, the preparation and processing of many defects, deformation difficult problem, as well as the common problem of the trade-off strength and ductility, the industrialization of magnesium alloys was limited. Therefore, machine learning can accelerate the discovery of novel magnesium alloys or processing parameters, and explore the relationship between their physicochemical characteristics and target properties. This paper comprehensively and systematically summarizes the research progress of machine learning in the applications of magnesium alloys. The basic process and various methods of machine learning are introduced, which mainly include dataset collection, data preprocessing, model building and performance evaluation. And the classification of machine learning algorithms is summarized briefly. Then, focusing on the research achievements of machine learning applied in many aspects such as magnesium alloys machining process, microstructure, mechanical properties, corrosion resistance, hydrogen storage properties, intrinsic properties (reinforcement mechanism, anisotropy, etc.) and inverse design. In particular, the alloy compositions, test temperature and time, the second phase, Schmid factor, and other various factors can be considered as features and input into machine learning models for training. The machine learning models not only accelerate novel high-performance magnesium alloy design but also provide an understanding of magnesium alloy mechanism research. Additionally, some urgently solved problems in the research and application of machine learning in magnesium alloys are analyzed. Such as the prediction of the chemical and physical properties of magnesium alloys is not enough, the prediction of the design and service performance of magnesium alloy components is just starting; and lacks the support of high-quality datasets. Finally, we propose future research directions and development trends in the application of machine learning in magnesium alloy.