Application of Computational Intelligence Methods for the Automated Identification of Paper-Ink Samples Based on LIBS

• DOI: 10.3390/s18113670
• Publication Date: 2018-10-29T15:10:41Z
• Authors: Krzysztof Rzecki; Tomasz Sośnicki; Mateusz Baran; Michał Niedźwiecki; Małgorzata Król; Tomasz Łojewski; U Rajendra Acharya; Özal Yildirim; Paweł Pławiak
• Keywords: Laser-induced breakdown spectroscopy; Perceptron; Multilayer perceptron; Probabilistic classification

Laser-induced breakdown spectroscopy (LIBS) is an important analysis technique with applications in many industrial branches and fields of scientific research. Nowadays, the advantages LIBS are impaired by main drawback interpretation obtained spectra identification observed spectral lines. This procedure highly time-consuming since it essentially based on comparison lines present spectrum literature database. paper proposes use various computational intelligence methods to develop a reliable fast classification quasi-destructively acquired into set predefined classes. We focus specific problem paper-ink samples 30 separate, For each classes (10 pens 5 ink types combined 10 sheets plus empty pages), 100 collected. Four variants preprocessing, seven classifiers (decision trees, random forest, k-nearest neighbor, support vector machine, probabilistic neural network, multi-layer perceptron, generalized regression network), 5-fold stratified cross-validation, test independent (for evaluation) scenarios employed. Our developed system yielded accuracy 99.08%, using forest classifier. results clearly demonstrates that machine learning can be used identify reliably at faster rate.