This chapter introduces two relatively recent regression methodologies based on a set of so‐called ‘natural computation’ methods. In contrast to classical programming, they work with rules rather than with well‐defined and fixed algorithms (like those presented in the previous chapters) and came into use during the 1990s and the 2000s.

Of the many natural computation techniques, artificial neural networks (ANNs) stand out, particularly their application in carrying out multivariate regression. As they constitute a promising way to cope with complex spectral problems (both molecular and atomic), they are introduced here. After an initial starting ‘rush’ they are now applied with some moderation and, mainly, to handle complex datasets (e.g. spectra with strong non‐linearities).