[19p-C601-1~15] 23.1 Joint Session N "Informatics"

3:15 PM - 3:30 PM JST(6:15 AM - 6:30 AM UTC)

[19p-C601-7] New material search maps corresponding to physical properties by machine learning

〇Yuki Inada¹, Erina Fujita², Kaoru Kimura², Yukari Katsura² (1.Univ. of Tokyo, 2.NIMS)

Keywords:

Materials Informatics,Machine learning,Materials Searching

The combinations of elements that can produce new materials for semiconductors, thermoelectric materials, superconductive materials, and quasicrystals were predicted by using machine learning. Focusing on combinations of elements within a ternary system, machine learning was performed with the combinations of elements that have known compounds showing each property as positive data, and the other combinations as unlabeled data. Using these machine learning models, the prediction results for all combinations of elements within the ternary system were shown on the heatmaps.