Detail publikačního výsledku

Interpreting convolutional neural network classifiers applied to laser-induced breakdown optical emission spectra

Kepes, E.; Vrábel, J.; Brázdil, T.; Holub, P.; Porízka, P.; Kaiser, J.

Originální název

Interpreting convolutional neural network classifiers applied to laser-induced breakdown optical emission spectra

Anglický název

Interpreting convolutional neural network classifiers applied to laser-induced breakdown optical emission spectra

Druh

Článek WoS

Originální abstrakt

Laser-induced breakdown spectroscopy (LIBS) is a well-established industrial tool with emerging relevance in high-stakes applications. To achieve its required analytical performance, LIBS is often coupled with advanced pattern-recognition algorithms, including machine learning models. Namely, artificial neural networks (ANNs) have recently become a frequently applied part of LIBS practitioners' toolkit. Nevertheless, ANNs are generally applied in spectroscopy as black-box models, without a real insight into their predictions. Here, we apply various post-hoc interpretation techniques with the aim of understanding the decision-making of convolutional neural networks. Namely, we find synthetic spectra that yield perfect expected classification predictions and denote these spectra class-specific prototype spectra. We investigate the simplest possible convolutional neural network (consisting of a single convolutional and fully connected layers) trained to classify the extended calibration dataset collected for the ChemCam laser-induced breakdown spectroscopy instrument of the Curiosity Mars rover. The trained convolutional neural network predominantly learned meaningful spectroscopic features which correspond to the elements comprising the major oxides found in the calibration targets. In addition, the discrete convolution operation with the learnt filters results in a crude baseline correction.

Anglický abstrakt

Laser-induced breakdown spectroscopy (LIBS) is a well-established industrial tool with emerging relevance in high-stakes applications. To achieve its required analytical performance, LIBS is often coupled with advanced pattern-recognition algorithms, including machine learning models. Namely, artificial neural networks (ANNs) have recently become a frequently applied part of LIBS practitioners' toolkit. Nevertheless, ANNs are generally applied in spectroscopy as black-box models, without a real insight into their predictions. Here, we apply various post-hoc interpretation techniques with the aim of understanding the decision-making of convolutional neural networks. Namely, we find synthetic spectra that yield perfect expected classification predictions and denote these spectra class-specific prototype spectra. We investigate the simplest possible convolutional neural network (consisting of a single convolutional and fully connected layers) trained to classify the extended calibration dataset collected for the ChemCam laser-induced breakdown spectroscopy instrument of the Curiosity Mars rover. The trained convolutional neural network predominantly learned meaningful spectroscopic features which correspond to the elements comprising the major oxides found in the calibration targets. In addition, the discrete convolution operation with the learnt filters results in a crude baseline correction.

Klíčová slova

Laser-induced breakdown spectroscopy; Classification; Interpretable machine learning; Convolutional neural networks; ChemCam calibration dataset

Klíčová slova v angličtině

Laser-induced breakdown spectroscopy; Classification; Interpretable machine learning; Convolutional neural networks; ChemCam calibration dataset

Autoři

Kepes, E.; Vrábel, J.; Brázdil, T.; Holub, P.; Porízka, P.; Kaiser, J.

Rok RIV

2025

Vydáno

01.01.2024

Nakladatel

ELSEVIER

Místo

AMSTERDAM

ISSN

1873-3573

Periodikum

TALANTA

Svazek

266

Číslo

1

Stát

Spojené království Velké Británie a Severního Irska

Strany počet

11

URL

BibTex

@article{BUT194128,
  author="Erik {Képeš} and Jakub {Vrábel} and Pavel {Pořízka} and Jozef {Kaiser} and Tomáš {Brázdil} and Petr {Holub}",
  title="Interpreting convolutional neural network classifiers applied to laser-induced breakdown optical emission spectra",
  journal="TALANTA",
  year="2024",
  volume="266",
  number="1",
  pages="11",
  doi="10.1016/j.talanta.2023.124946",
  issn="0039-9140",
  url="https://www-sciencedirect-com.ezproxy.lib.vutbr.cz/science/article/pii/S0039914023006975?via%3Dihub"
}