For full functionality of this site it is necessary to enable JavaScript. Here are the instructions how to enable JavaScript in your web browser.
SIMS21, Poland 2017 - Alex Henderson abstract

Alex Henderson oral presentation (FN2-Tue2-1-5)

Rise of the Machines: The Use of Machine Learning in SIMS Data Analysis

Alex Henderson1,2

1 University of Manchester - Manchester Institute of Biotechnology, 131 Princess Street, M1 7DN Manchester, United Kingdom
2 SurfaceSpectra Ltd., 32 Bury Avenue, M16 0AT Manchester, United Kingdom


The huge amount of data, encoded in anything but the most simplistic SIMS spectrum, requires the use of multivariate analysis techniques to extract meaningful information. There has been a long history of SIMS practitioners using parametric statistical approaches, such as principal components analysis (PCA), to simplify their data. Taken with a priori knowledge, discriminant analysis methods can be used to good effect in order to classify samples via their spectra. In some cases these can also determine the spectral responses that give rise to the classification requested. However, these approaches rely on an underlying assumption that the data is normally distributed; an assumption that is difficult to prove.

In recent years, the power and ease-of-use of computers has greatly increased. Data analysis approaches, that would have been prohibitively time consuming in the past, are now possible. This brings another, non-parametric, approach to the table: Machine Learning.

In this presentation we will cover the differences between parametric and non-parametric approaches. Using example data, we will show how machine learning tools can be used to interrogate static SIMS spectra, the types of output available and the limitations, or otherwise, of various methods.