Abstract: Porous materials have attracted much attention due to their wide potential applications. Traditionally, research on the mechanical properties of porous materials has mainly relied on time-consuming and cumbersome experimental and theoretical analysis methods. In recent years, machine learning technology has provided an efficient solution to simplify the complex relationship between porous material parameters and mechanical properties. This article reviews the latest research progress of machine learning in predicting the mechanical properties of porous materials. Firstly, commonly used machine learning algorithms were introduced, with a focus on analyzing the application of neural network prediction methods in this field. This method was summarized into three major strategies: mechanism model driven neural network, integration of neural networks and mechanistic model, and integration of neural network and optimization technology. Then, the basic principles and applications of the above strategies were analyzed in detail. Finally, we discussed how to develop more efficient hybrid models by improving neural network technology and integrating it with optimization algorithms, and looked forward to the development prospects of neural networks in this field.