Prediction of non-measured time spectra from measured one by Random Forest and Kolmogorov Arnold Network (KAN) regressors

The machine-learning (ML) technique is a strong tool to handle big data. In the future the SuperMuSR spectrometer will give us huge amount of data rather than now making SR data analysis complicate and harder. We are developing ML data analysis method by using Random Forest and Kolmogorov Arnold Network regressors to predict non-measured time spectra from the measured one in order to realize a ML-controlled SR data taking and analysis. For this purpose, we propose to measure the high-purity Cu and obtained reference data which can be used to train those ML regressors.