Treffer: High-entropy sulfide data for clustering comparison

Unsupervised machine learning algorithms are applied to two different sets of scanning transmission electron microscope data. Energy dispersive X-ray (EDX) analyses were performed on two different samples, a bulk 6-element sulfide (in the file 0008 - SI HAADF 4000 x Nano.hspy) and a nanoparticulate 7-element sulfide (in file HE03NP1F1.emd). Unsupervised clustering was performed using a gaussian mixture model (GMM) and by the HDBSCAN algorithm. This allowed the dataset to be segmented to help automatically identify phase and composition variations within samples with minimal user input. The python notebook can be used to open and perform the clustering analysis on both of the attached datatasets.