Using machine learning methods to predict the transition temperature in superconducting materials

Gashmard, H., Shakeripour, H.

Using machine learning methods to predict the transition temperature in superconducting materials

(2022) 6 July. 7^th National Conference on Progress in Superconductivity and Magnetism, Iran. p. 91-94.

Abstract

One of the methods for determining the properties of materials is the use of simulation and computational methods. Today, the impressive results of artificial intelligence approaches and machine learning algorithms have convinced condensed-matter researchers to solve their advancement problems in this way. In this study, using the CatBoost algorithm, the transition temperature of 1286 superconductors was predicted, and determined that the average error for this number of superconductors was 6.19 K and R 2 was 0.956. Twentynine properties were considered for each superconducting compound.

Keywords: Superconducting Materials, Machine Learning, CatBoost algorithm