石油学会 年会（研究発表会）

Materials informatics(MI), including machine learning methods for material design, is expected to be increasingly important as an essential methodology for catalyst development, with various materials databases. In the early stages of MI, I have primary utilized support vector machine (SVM) and genetic algorithms for catalyst design. They are robust methods to improve catalyst activity, selectivity, and life by optimizing catalyst components, catalyst preparation conditions, and reaction conditions. SVM can also be used as a tool to estimate the key factors that determine catalytic activity.