[22a-M206-1~11] 23.1 Joint Session N "Informatics"

△ [22a-M206-4] Synthesizability Prediction of Ternary Compounds by Machine Learning

〇Yuki Inada¹, Kaoru Kimura², Yukari Katsura² (1.Univ. Tokyo, 2.NIMS)

Keywords:Materials Informatics, Machine learning

We tried to predict the compositions of stable compounds by machine learning for new materials search. Training data was about 20,000 ternary stable materials from Materials Project. Machine learning models were trained to predict whether input compositions were included in the training data. Using these machine learning models, prediction results were shown on the ternary diagrams of over 80,000 combinations of elements.