EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Adjoint propagation of error signal through modular recurrent neural networks for biologically plausible learning

Zhuo Liu iD, Hao Shu iD, Linmiao Wang, Xu Meng, Yousheng Wang, Xuancheng Li, Wei Wang, Tao Chen iD

eLife · 2026

DOI: 10.7554/elife.108237 ↗

Kurz gefasst

Fachartikel in der Zeitschrift eLife, 2026. Verfasst von Zhuo Liu und 7 weiteren Autor:innen. Beteiligte Einrichtung: School of Microelectronics, University of Science and Technology of China, Hefei, China. Hauptthemen laut MEDLINE: Neural Networks, Computer, Brain, Learning. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 93 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Chinese Academy of Sciences, University of Science and Technology of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Biologically plausible learning mechanisms have implications for understanding brain functions and engineering intelligent systems. Inspired by the multi-scale recurrent connectivity in the brain, we introduce an adjoint propagation (AP) framework, in which the error signals arise naturally from recurrent dynamics and propagate concurrently with forward inference signals. AP inherits the modularity of multi-region recurrent neural network (MR-RNN) models and leverages the convergence properties of RNN modules to facilitate fast and scalable training. This framework eliminates the biologically implausible feedback required by the backpropagation (BP) algorithm and allows concomitant error propagation for multiple tasks through the same RNN. We demonstrate that AP succeeds in training on standard benchmark tasks, achieving accuracies comparable to BP-trained networks while adhering to neurobiological constraints. The training process exhibits robustness, maintaining performance over extended training epochs. Importantly, AP supports flexible resource allocation for multiple cognitive tasks, consistent with observations in neuroscience. This framework bridges artificial and biological learning principles, paving the way for energy-efficient intelligent systems inspired by the brain and offering a mechanistic theory that can guide experimental investigations in neuroscience.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Neural Networks, ComputerBrainLearningAlgorithmsRecurrent Neural NetworksHumansSoft ComputingModels, Neurological

Stichwörter der Autor:innen

artificial neuroncomputer algorithmneurosciencerecurrent neural network

Autor:innen und Institutionen

  • School of Microelectronics, University of Science and Technology of China, Hefei, China.
  • Digital Intelligence Centre, Shenzhen Power Supply Bureau, China Southern Power Grid, Shenzhen, China.

ORCID-Profile: Zhuo Liu, Hao Shu, Tao Chen

Förderung

  • Chinese Academy of Sciences
  • University of Science and Technology of China

Zitieren

APA

Liu, Z., Shu, H., Wang, L., Meng, X., Wang, Y., Li, X., Wang, W., & Chen, T. (2026). Adjoint propagation of error signal through modular recurrent neural networks for biologically plausible learning. eLife. https://doi.org/10.7554/elife.108237

Vancouver

Liu Z, Shu H, Wang L, Meng X, Wang Y, Li X, et al. Adjoint propagation of error signal through modular recurrent neural networks for biologically plausible learning. eLife. 2026. doi:10.7554/elife.108237

BibTeX
@article{Liu2026,
  author = {Liu, Zhuo and Shu, Hao and Wang, Linmiao and Meng, Xu and Wang, Yousheng and Li, Xuancheng and Wang, Wei and Chen, Tao},
  title = {Adjoint propagation of error signal through modular recurrent neural networks for biologically plausible learning},
  journal = {eLife},
  year = {2026},
  doi = {10.7554/elife.108237}
}
Kostenlos starten →