Ab initio machine learning for accelerating catalytic materials discovery

Machine learning, a branch of artificial intelligence techniques for generalized prediction, has recently shown great promise in accelerating catalytic materials discovery when combined with ab initio calculations. In this chapter, machine learning models and their applications in catalyst design with an emphasis on heterogeneous catalysis are reviewed. Examples of recent research work with various algorithms, features, and learning strategies are provided for readers to better understand this area. In addition, machine learning in homogeneous catalysis is also briefly introduced with case study examples. Finally, challenges and opportunities are discussed, and we believe that such outlooks will be helpful for researchers who are entering this rapidly emerging field.