Data-driven exploration of new superconductors under high pressure

10:35-11:05

Development of high quality semiconductor crystal growth process using machine learning.

11:05-11:35

Development of Ultranarrow-Band Thermal Emission with Multilayers Designed by Machine Learning

11:35-12:05

Material design of filled rubbers using machine learning

12:05-12:35

【Profile】

Dr. Koishi joined Yokohama Rubber Company in 1985. After serving as the head of the CAE laboratory and the head of the KOISHI laboratory at the company, he has been the associate corporate officer of the company and the head of the KOISHI laboratory since 2014. Since joining the company, he has been in charge of research on computational mechanics, multi-objective optimization, data mining and machine learning for tires and rubber materials. He is also a fellow of the Japan Society of Mechanical Engineers, and chairman of the 92nd Computational Mechanics Division of the Society. He was praised for his achievements in the field of computational mechanics and received the Computational Mechanics Award from the Society in 2016. Currently he is the chairman of the Kanto CAE Konwakai.

【Abstract】

Focusing on design and development in the manufacturing industry, we expect data-driven "understanding" and "discovery" for machine learning. Searching outside the conventional domain i.e. extrapolated search is essential for new discoveries, so machine learning based on test data is often insufficient. One solution is a combination of computational science and machine learning. In this lecture, materials informatics for rubber material design that we have been working on so far will be presented with personal opinions.