LAUSR

With the coming of the third age of artificial intelligence, machine learning has been successfully implemented in many fields, which changed the paradigms to feed tons of data to find the pattern that with similar features. However, limited studies are focused on the electron microscopy and corresponding spectrum applications. In this study, we develop a novel de-noising approach based on the clustering algorithm, named kMMLS, which combines the k-means clustering and the multiple linear least squares (MLLS). The kMLLS clustering routine can extract the nearly pure endmember and be applied in the region of Electron Energy Loss Spectroscopy (EELS) investigations. With using the extracted endmember as the reference spectra of MLLS fitting, we can obtain the de-noising data set from the MLLS routine.