Chapter 6: Machine Learning in Porous Materials

In this chapter, recent advances in machine learning are discussed as they relate to computational achievements for porous materials. Machine learning has accelerated simulations for property calculation, screening for desired chemical behaviour, and for the informed synthesis of porous materials. The use of machine learning models has value in the design and synthesis of porous materials because of the large, structured, combinatorial space that is computationally intensive to traverse with incremental simulation approaches. This has motivated the use of machine learning methods, but there are particular challenges with the application of these methods that must be overcome, such as the development of appropriate input descriptors, so-called feature vectors, data availability and interoperability, as well as method development.