The machine learning models for prediction superconducting properties of materials
Description: My research is based on three main methods of machine learning - linear regression, random forest, neural network. Also, it was created a graphical interface for temperature prediction transition of superconductors based on their basic properties. This will allow you to open new materials faster.
Superconductors can perform technological breakthroughs in many fields, particularly in energy, electronics, and quantum computers. However, many problems, both technological and scientific, remained unsolved. If it is possible to combine two different fields – computer science and physics? Obviously, yes. Nowadays, we have the century of information technology which not only makes our life easier but also helps to investigate new materials. Due to that, I decided to find the solution of high-temperature superconductors, predict the critical temperature of compounds using machine learning. My research is based on three main methods of machine learning- linear regression, random forest and neural network. At first, the analysis was done together with the all data. Also, it was created a graphical interface that uses the main results of neural network and Random Forest for temperature prediction transition of superconductors based on their basic properties. Other substances that are not yet known as superconductors can also be tested on them. This will make it possible to obtain an approximate transition temperature and experimentally obtain new superconducting compounds.
Organisation: National Center “Junior Academy of Sciences of Ukraine” under the auspices of UNESCO
Innovator(s): Anastasiia Veretilnyk
Category: Other/Miscellaneous
Country: Ukraine
