Hybrid Variable Selection and Support Vector Regression for Gas Sensor Optimization

Margarita Rebolledo*, Ruxandra Stoean, A. E. Eiben, Thomas Bartz-Beielstein

*Corresponding author for this work

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

The improvement of combustion processes in industry, especially in the automotive branch, is of great importance to maintain the environmental permitted limits. Carbon monoxide concentration in the exhaust gases can give an insight into the efficiency of the combustion taking place and for this reason, it is important to have sensors that can measure it accurately. First results of a long term study with one of the leading sensor manufactures showed high performance using genetic programming. However, this expensive approach is difficult to apply in real-world settings. Therefore a hybrid optimization that combines support vector regression (SVR) with variable pre-selection is proposed. Three different methods for variable selection are compared for this application, a genetic algorithm, and two methods from Bayesian statistics: statistical equivalent signatures and projection predictive variable selection. Furthermore, a multi-objective approach using the same hybrid definition is implemented for the cases in which several sensors need to be considered simultaneously. Our results show that the hybrid model is an improvement compared to the previous study, while delivering good performance when dealing with a multivariate formulation. Genetic algorithms in combination with SVR lead to enhanced variation on the groups of selected variables.

Original languageEnglish
Title of host publicationBioinspired Optimization Methods and Their Applications
Subtitle of host publication9th International Conference, BIOMA 2020 Brussels, Belgium, November 19–20, 2020 Proceedings
EditorsBogdan Filipic, Edmondo Minisci, Massimiliano Vasile
PublisherSpringer Science and Business Media Deutschland GmbH
Pages281-293
Number of pages13
ISBN (Electronic)9783030637101
ISBN (Print)9783030637095
DOIs
Publication statusPublished - 2020
Event9th International Conference on Bioinspired Optimization Methods and Their Applications, BIOMA 2020 - Brussels, Belgium
Duration: 19 Nov 202020 Nov 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12438 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference9th International Conference on Bioinspired Optimization Methods and Their Applications, BIOMA 2020
CountryBelgium
CityBrussels
Period19/11/2020/11/20

Keywords

  • Feature selection
  • Projection predictive
  • Statistical equivalent signatures
  • Support vector regression

Fingerprint Dive into the research topics of 'Hybrid Variable Selection and Support Vector Regression for Gas Sensor Optimization'. Together they form a unique fingerprint.

Cite this