.png){kind=icon}
.png){kind=icon}
Return
Accelerating the search for superconductors using machine learning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.commatsci.2025.114453.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A refined and validated superconducting materials dataset addressing prior inconsistencies. • Principal Component Analysis showing that QSD-based descriptors effectively cluster the superconductors classwise in the phase space. • Established the physical interpretability of the chosen descriptors to Tc using SHAP analysis. • Validation: - (a) Prediction of critical temperatures of 27 recently reported compounds that are outside the dataset used in training here. - (b) Identification of 21 promising superconductor materials from large material databases such as Materials Project, 8 of which have been reported for superconductivity. • The database, trained model, and the Python scripts used for model training, descriptor generation and Tc prediction in this project have been made accessible through a GitHub repository, adhering to the FAIR (Findable, Accessible, Interoperable, Reusable) principles to ensure transparency, reproducibility, and community reuse.
Journal
IF:
3.3
Papers:
1.3W
Citations:
3.6W