Non-linear shrinking of linear model errors.

Deep learning Hybrid model Interpretation Neural network PLSR Residual modelling

Journal

Analytica chimica acta
ISSN: 1873-4324
Titre abrégé: Anal Chim Acta
Pays: Netherlands
ID NLM: 0370534

Informations de publication

Date de publication:
01 Jun 2023
Historique:
received: 21 12 2022
revised: 17 03 2023
accepted: 24 03 2023
medline: 23 4 2023
pubmed: 23 4 2023
entrez: 22 04 2023
Statut: ppublish

Résumé

Artificial neural networks (ANNs) can be a powerful tool for spectroscopic data analysis. Their ability to detect and model complex relations in the data may lead to outstanding predictive capabilities, but the predictions themselves are difficult to interpret due to the lack of understanding of the black box ANN models. ANNs and linear methods can be combined by first fitting a linear model to the data followed by a non-linear fitting of the linear model residuals using an ANN. This paper explores the use of residual modelling in high-dimensional data using modern neural network architectures. By combining linear- and ANN modelling, we demonstrate that it is possible to achieve both good model performance while retaining interpretations from the linear part of the model. The proposed residual modelling approach is evaluated on four high-dimensional datasets, representing two regression and two classification problems. Additionally, a demonstration of possible interpretation techniques are included for all datasets. The study concludes that if the modelling problem contains sufficiently complex data (i.e., non-linearities), the residual modelling can in fact improve the performance of a linear model and achieve similar performance as pure ANN models while retaining valuable interpretations for a large proportion of the variance accounted for. The paper presents a residual modelling scheme using modern neural network architectures. Furthermore, two novel extensions of residual modelling for classification tasks are proposed. The study is seen as a step towards explainable AI, with the aim of making data modelling using artificial neural networks more transparent.

Sections du résumé

BACKGROUND BACKGROUND
Artificial neural networks (ANNs) can be a powerful tool for spectroscopic data analysis. Their ability to detect and model complex relations in the data may lead to outstanding predictive capabilities, but the predictions themselves are difficult to interpret due to the lack of understanding of the black box ANN models. ANNs and linear methods can be combined by first fitting a linear model to the data followed by a non-linear fitting of the linear model residuals using an ANN. This paper explores the use of residual modelling in high-dimensional data using modern neural network architectures.
RESULTS RESULTS
By combining linear- and ANN modelling, we demonstrate that it is possible to achieve both good model performance while retaining interpretations from the linear part of the model. The proposed residual modelling approach is evaluated on four high-dimensional datasets, representing two regression and two classification problems. Additionally, a demonstration of possible interpretation techniques are included for all datasets. The study concludes that if the modelling problem contains sufficiently complex data (i.e., non-linearities), the residual modelling can in fact improve the performance of a linear model and achieve similar performance as pure ANN models while retaining valuable interpretations for a large proportion of the variance accounted for.
SIGNIFICANCE AND NOVELTY UNASSIGNED
The paper presents a residual modelling scheme using modern neural network architectures. Furthermore, two novel extensions of residual modelling for classification tasks are proposed. The study is seen as a step towards explainable AI, with the aim of making data modelling using artificial neural networks more transparent.

Identifiants

pubmed: 37087289
pii: S0003-2670(23)00368-9
doi: 10.1016/j.aca.2023.341147
pii:
doi:

Types de publication

Journal Article

Langues

eng

Sous-ensembles de citation

IM

Pagination

341147

Informations de copyright

Copyright © 2023 The Author(s). Published by Elsevier B.V. All rights reserved.

Déclaration de conflit d'intérêts

Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Auteurs

Runar Helin (R)

Norwegian University of Life Sciences, Faculty of Science and Technology, Ås, Norway. Electronic address: runar.helin@nmbu.no.

Ulf Indahl (U)

Norwegian University of Life Sciences, Faculty of Science and Technology, Ås, Norway.

Oliver Tomic (O)

Norwegian University of Life Sciences, Faculty of Science and Technology, Ås, Norway.

Kristian Hovde Liland (KH)

Norwegian University of Life Sciences, Faculty of Science and Technology, Ås, Norway.

Classifications MeSH