Physics-guided NMF for STEM/EDXS Data Analysis

This lecture covers the application of Physics-guided Non-negative Matrix Factorization (NMF) for the analysis of Scanning Transmission Electron Microscopy (STEM) and Energy-Dispersive X-ray Spectroscopy (EDXS) data. It explains the challenges of noisy data and the linear mixing model, leading to the introduction of NMF to solve the mixing problem. The lecture delves into the MLATEM project, the optimization process for learning W and H, constraints, and regularizations. It also discusses the advantages of the modelization, phase separation, quantification, and the use of simulated data for testing different algorithms.