Preprocessing of structured spectral data to improve the predictive accuracy of Self-Organising Maps
Preprocessing of structured spectral data to improve the predictive accuracy of Self-Organising Maps
Domaschke, K.; Rossberg, A.; Villmann, T.
In this paper, we propose a new approach using structural information of spectral data during a preprocessing procedure to upgrade the ability of subsequent analysis methods. A composite data set of measured spectra is given, which contains dierent mixtures of a few spectral components. Using chemical knowledge and a small subset of the mixture information, we are able to evaluate these spectral components out of the given data set and use this information in addition for the following analysis of the composite data set. In our case, we apply the Self-Organizing Map according to Kohonen to predict the unknown mixture subset of the dierent spectral components within the measured data.
Keywords: Self-Organizing maps; spectral data; unmixing; blind source separation
Involved research facilities
- Rossendorf Beamline at ESRF DOI: 10.1107/S1600577520014265
Related publications
- DOI: 10.1107/S1600577520014265 is cited by this (Id 19450) publication
-
Contribution to proceedings
European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, 23.-25.04.2014, Bruges, Belgium
Proceedings of the European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Permalink: https://www.hzdr.de/publications/Publ-19450