Journal cover Journal topic
Geoscientific Instrumentation, Methods and Data Systems An interactive open-access journal of the European Geosciences Union
Journal topic

Journal metrics

Journal metrics

  • IF value: 1.302 IF 1.302
  • IF 5-year value: 1.356 IF 5-year
    1.356
  • CiteScore value: 1.47 CiteScore
    1.47
  • SNIP value: 0.810 SNIP 0.810
  • IPP value: 1.37 IPP 1.37
  • SJR value: 0.598 SJR 0.598
  • Scimago H <br class='hide-on-tablet hide-on-mobile'>index value: 8 Scimago H
    index 8
  • h5-index value: 14 h5-index 14
Discussion papers
https://doi.org/10.5194/gi-2019-41
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/gi-2019-41
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Submitted as: research article 28 Jan 2020

Submitted as: research article | 28 Jan 2020

Review status
This preprint is currently under review for the journal GI.

Auroral Classification Ergonomics and the Implications for Machine Learning

Derek McKay1 and Andreas Kvammen2 Derek McKay and Andreas Kvammen
  • 1NORCE Norwegian Research Centre AS, Tromsø, Norway
  • 2Department of Physics and Technology, UiT – The Arctic University of Norway, Tromsø, Norway

Abstract. The machine learning research community has focused greatly on bias in algorithms and have identified different manifestations of it. Bias in the training samples is recognised as a potential source of prejudice in machine learning. It can be introduced by human experts who define the training sets. As machine learning techniques are being applied to auroral classification, it is important to identify and address potential sources of expert-injected bias. In an ongoing study, 13 947 auroral images were manually classified with significant differences between classifications. This large data set allowed identification of some of these biases, especially those originating as a result of the ergonomics of the classification process. These findings are presented in this paper, to serve as a checklist for improving training data integrity, not just for expert classifications, but also for crowd-sourced, citizen science projects. As the application of machine learning techniques to auroral research is relatively new, it is important that biases are identified and addressed before they become endemic in the corpus of training data.

Derek McKay and Andreas Kvammen

Interactive discussion

Status: open (until 09 Apr 2020)
Status: open (until 09 Apr 2020)
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
[Subscribe to comment alert] Printer-friendly Version - Printer-friendly version Supplement - Supplement

Derek McKay and Andreas Kvammen

Derek McKay and Andreas Kvammen

Viewed

Total article views: 219 (including HTML, PDF, and XML)
HTML PDF XML Total Supplement BibTeX EndNote
187 28 4 219 17 10 7
  • HTML: 187
  • PDF: 28
  • XML: 4
  • Total: 219
  • Supplement: 17
  • BibTeX: 10
  • EndNote: 7
Views and downloads (calculated since 28 Jan 2020)
Cumulative views and downloads (calculated since 28 Jan 2020)

Viewed (geographical distribution)

Total article views: 201 (including HTML, PDF, and XML) Thereof 200 with geography defined and 1 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 

Cited

Saved

No saved metrics found.

Discussed

No discussed metrics found.
Latest update: 06 Apr 2020
Publications Copernicus
Download
Short summary
Researchers are making increasing use of machine learning to improve accuracy, efficiency and consistency. During such a study of the aurora, it was noted that biases or distortions had crept into the data, because of the conditions (or ergonomics) of the human trainers. As using machine learning techniques to auroral research is relatively new, it is critical that such biases are brought to the attention of the academic and citizen science communities.
Researchers are making increasing use of machine learning to improve accuracy, efficiency and...
Citation