EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Image segmentation with traveling waves in an exactly solvable recurrent neural network

Luisa H B Liboni iD, Roberto C Budzinski iD, Alexandra N Busch, Sindy Löwe, Thomas A Keller, Max Welling iD, Lyle E Muller iD

Proceedings of the National Academy of Sciences · 07.01.2025 · 122 (1)

DOI: 10.1073/pnas.2321319121 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 03.01.2025. Verfasst von Luisa H B Liboni und 6 weiteren Autor:innen. Beteiligte Einrichtung: Department of Mathematics, Western University, London, ON N6A 3K7, Canada. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NEI NIH HHS, NINDS NIH HHS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

We study image segmentation using spatiotemporal dynamics in a recurrent neural network where the state of each unit is given by a complex number. We show that this network generates sophisticated spatiotemporal dynamics that can effectively divide an image into groups according to a scene’s structural characteristics. We then demonstrate a simple algorithm for object segmentation that generalizes across inputs ranging from simple geometric objects in grayscale images to natural images. Using an exact solution of the recurrent network’s dynamics, we present a precise description of the mechanism underlying object segmentation in the network dynamics, providing a clear mathematical interpretation of how the algorithm performs this task. Object segmentation across all images is accomplished with one recurrent neural network that has a single, fixed set of weights. This demonstrates the expressive potential of recurrent neural networks when constructed using a mathematical approach that brings together their structure, dynamics, and computation.

Schlagwörter

Stichwörter der Autor:innen

explainable AIimage segmentationrecurrent neural networksspatiotemporal dynamicsvisual system

Autor:innen und Institutionen

  • Department of Mathematics, Western University, London, ON N6A 3K7, Canada.
  • Western Institute for Neuroscience, Western University, London, ON N6A 3K7, Canada.
  • Western Academy for Advanced Research, Western University, London, ON N6A 3K7, Canada.
  • Fields Lab for Network Science, Fields Institute, Toronto, ON M5T 3J1, Canada.
  • Amsterdam Machine Learning Lab, University of Amsterdam, Amsterdam 1012 WP, The Netherlands.
  • University of Amsterdam-Bosch Deep Learning Technologies Amsterdam Lab, University of Amsterdam, Amsterdam 1012 WP, The Netherlands.

ORCID-Profile: Luisa H B Liboni, Roberto C Budzinski, Max Welling, Lyle E Muller

Förderung

  • NEI NIH HHS
  • NINDS NIH HHS

Zitieren

APA

Liboni, L. H. B., Budzinski, R. C., Busch, A. N., Löwe, S., Keller, T. A., Welling, M., & Muller, L. E. (2025). Image segmentation with traveling waves in an exactly solvable recurrent neural network. Proceedings of the National Academy of Sciences, 122(1). https://doi.org/10.1073/pnas.2321319121

Vancouver

Liboni LHB, Budzinski RC, Busch AN, Löwe S, Keller TA, Welling M, et al. Image segmentation with traveling waves in an exactly solvable recurrent neural network. Proceedings of the National Academy of Sciences. 2025;122(1). doi:10.1073/pnas.2321319121

BibTeX
@article{Liboni2025,
  author = {Liboni, Luisa H B and Budzinski, Roberto C and Busch, Alexandra N and Löwe, Sindy and Keller, Thomas A and Welling, Max and Muller, Lyle E},
  title = {Image segmentation with traveling waves in an exactly solvable recurrent neural network},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2025},
  volume = {122},
  number = {1},
  doi = {10.1073/pnas.2321319121}
}
Kostenlos starten →