Recently, materials informatics exploiting machine learning techniques to accelerate materials research has been developing rapidly. Under these circumstances, we have developed a method for high-througput spectroscopies that implements a basic technique of machine learning, Gaussian process regression (GPR). This method allows for prediction of detailed peak structures by sampling only about one-sixth of original sampling points in conventional spectroscopies.

Y. K. Wakabayashi*, et al.
Applied Physics Express 11, 112401 (2018).