EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Biologically plausible gated recurrent neural networks for working memory and learning-to-learn

Alexandra R van den Berg iD, Pieter R Roelfsema iD, Sander M Bohte iD

PLOS ONE · 2024

DOI: 10.1371/journal.pone.0316453 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS ONE, 2024. Verfasst von Alexandra R van den Berg und 2 weiteren Autor:innen. Beteiligte Einrichtung: Machine Learning Group, Centrum Wiskunde & Informatica, Amsterdam, The Netherlands. Hauptthemen laut MEDLINE: Memory, Short-Term, Learning, Neural Networks, Computer. Die Arbeit nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: HORIZON EUROPE Framework Programme, Agence Nationale de la Recherche (ANR) within Programme Investissements d’Avenir, Institut Hospitalo-Universitaire FOReSIGHT, Ministerie van Onderwijs, Cultuur en Wetenschap.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The acquisition of knowledge and skills does not occur in isolation but learning experiences amalgamate within and across domains. The process through which learning can accelerate over time is referred to as learning-to-learn or meta-learning. While meta-learning can be implemented in recurrent neural networks, these networks tend to be trained with architectures that are not easily interpretable or mappable to the brain and with learning rules that are biologically implausible. Specifically, these rules have often employed backpropagation-through-time, which relies on information that is unavailable at synapses that are undergoing plasticity in the brain. Previous studies that exclusively used local information for their weight updates had a limited capacity to integrate information over long timespans and could not easily learn-to-learn. Here, we propose a novel gated memory network named RECOLLECT, which can flexibly retain or forget information by means of a single memory gate and is trained with a biologically plausible trial-and-error-learning that requires only local information. We demonstrate that RECOLLECT successfully learns to represent task-relevant information over increasingly long memory delays in a pro-/anti-saccade task, and that it learns to flush its memory at the end of a trial. Moreover, we show that RECOLLECT can learn-to-learn an effective policy on a reversal bandit task. Finally, we show that the solutions acquired by RECOLLECT resemble how animals learn similar tasks.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Memory, Short-TermLearningNeural Networks, ComputerHumansModels, NeurologicalAnimalsBrainNerve Net

Autor:innen und Institutionen

  • Machine Learning Group, Centrum Wiskunde & Informatica, Amsterdam, The Netherlands.
  • Department of Vision & Cognition, Netherlands Institute for Neuroscience, Amsterdam, The Netherlands.
  • Department of Integrative Neurophysiology, Center for Neurogenomics and Cognitive Research, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
  • Department of Neurosurgery, Academic Medical Center, Amsterdam, The Netherlands.
  • Laboratory of Visual Brain Therapy, Institut National de la Santé et de la Recherche Médicale, Institut de la Vision, Sorbonne Université, Centre National de la Recherche Scientifique, Paris, France.
  • Swammerdam Institute of Life Sciences, University of Amsterdam, Amsterdam, The Netherlands.

ORCID-Profile: Alexandra R van den Berg, Pieter R Roelfsema, Sander M Bohte

Förderung

  • HORIZON EUROPE Framework Programme
  • Agence Nationale de la Recherche (ANR) within Programme Investissements d’Avenir, Institut Hospitalo-Universitaire FOReSIGHT
  • Ministerie van Onderwijs, Cultuur en Wetenschap

Zitieren

APA

van den Berg, A. R., Roelfsema, P. R., & Bohte, S. M. (2024). Biologically plausible gated recurrent neural networks for working memory and learning-to-learn. PLOS ONE. https://doi.org/10.1371/journal.pone.0316453

Vancouver

van den Berg AR, Roelfsema PR, Bohte SM. Biologically plausible gated recurrent neural networks for working memory and learning-to-learn. PLOS ONE. 2024. doi:10.1371/journal.pone.0316453

BibTeX
@article{vandenBerg2024,
  author = {van den Berg, Alexandra R and Roelfsema, Pieter R and Bohte, Sander M},
  title = {Biologically plausible gated recurrent neural networks for working memory and learning-to-learn},
  journal = {PLOS ONE},
  year = {2024},
  doi = {10.1371/journal.pone.0316453}
}
Kostenlos starten →