Chapter 17: Artificial Intelligence, Machine Learning, and XRF

The application of machine learning and artificial intelligence has the potential to expand the capabilities of X-ray fluorescence analysis. This includes improving mastering the identification of elements for which XRF is already applicable, identifying lower determinable quantities, and even estimating properties of the sample which do not have a characteristic K, L, or M emission line. Machine learning calibrations, despite their potential, are still limited by the same constraints common across empirical methods; they do not extrapolate well beyond the standards used to inform them. The present chapter examines multiple machine learning architectures for XRF calibrations, with a discussion of their strengths and weaknesses. In addition, a number of examples of both quantitative and qualitative machine learning models are used to highlight the flexibility of the approach.