Presentation Information
[20a-P04-3]Machine Learning Workflow for Discovering New Superconductors
〇Harunobu Ozawa1, Kakeru Matsushita1, Masatomo Uehara1 (1.YNU)
Keywords:
superconductor,superconductivity
Many attempts using AI have been made around the world to predict the new superconductors. We have attempted to develop a workflow to discover new superconductors efficiently and accurately. For this purpose we have combined a model that predicts the candidates for new superconductors using a dimensionality reduction algorithm called UMAP and a deep learning model that predicts transition temperatures from the crystal structures of predicted new superconductor candidates.