Interpretation of material spectra can be data-driven using machine learning

Modern spectroscopy techniques can produce tens of thousands of spectra from a single experiment, which has placed a considerable burden on traditional human-driven methods for interpretation of these spectra. A research team combined two machine learning techniques, layer clustering and decision tree methods, to produce data-driven methods for spectral interpretation and prediction that can analyze any spectral data quickly and accurately.

from Engineering and Construction News – ScienceDaily https://www.sciencedaily.com/releases/2018/09/180906082027.htm

via Tumblr http://ndbasilica.tumblr.com/post/177841246574