EUVIMEDEUVIMED entdecken
Open Access

A brain-inspired algorithm that mitigates catastrophic forgetting of artificial and spiking neural networks with low computational cost

Tielin Zhang iD, Xiang Cheng iD, Shuncheng Jia iD, Chengyu T Li iD, Mu-Ming Poo, Bo Xu iD

Science Advances · 25.08.2023 · 9 (34)

DOI: 10.1126/sciadv.adi2947 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science Advances, erschienen am 25.08.2023. Verfasst von Tielin Zhang und 5 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China. Hauptthemen laut MEDLINE: Algorithms, Neural Networks, Computer. Die Arbeit wurde bisher 39-mal zitiert (Crossref) und nennt 52 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Neuromodulators in the brain act globally at many forms of synaptic plasticity, represented as metaplasticity, which is rarely considered by existing spiking (SNNs) and nonspiking artificial neural networks (ANNs). Here, we report an efficient brain-inspired computing algorithm for SNNs and ANNs, referred to here as neuromodulation-assisted credit assignment (NACA), which uses expectation signals to induce defined levels of neuromodulators to selective synapses, whereby the long-term synaptic potentiation and depression are modified in a nonlinear manner depending on the neuromodulator level. The NACA algorithm achieved high recognition accuracy with substantially reduced computational cost in learning spatial and temporal classification tasks. Notably, NACA was also verified as efficient for learning five different class continuous learning tasks with varying degrees of complexity, exhibiting a markedly mitigated catastrophic forgetting at low computational cost. Mapping synaptic weight changes showed that these benefits could be explained by the sparse and targeted synaptic modifications attributed to expectation-based global neuromodulation.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

AlgorithmsNeural Networks, ComputerBrainLearningRecognition, Psychology

Autor:innen und Institutionen

  • Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China.
  • School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China.
  • Shanghai Center for Brain Science and Brain-inspired Technology, Lingang Laboratory, Shanghai 200031, China.
  • Center for Excellence in Brain Science and Intelligence Technology, Institute of Neuroscience, Chinese Academy of Sciences, Shanghai 200031, China.

ORCID-Profile: Tielin Zhang, Xiang Cheng, Shuncheng Jia, Chengyu T Li, Bo Xu

Zitieren

APA

Zhang, T., Cheng, X., Jia, S., Li, C. T., Poo, M. M., & Xu, B. (2023). A brain-inspired algorithm that mitigates catastrophic forgetting of artificial and spiking neural networks with low computational cost. Science Advances, 9(34). https://doi.org/10.1126/sciadv.adi2947

Vancouver

Zhang T, Cheng X, Jia S, Li CT, Poo MM, Xu B. A brain-inspired algorithm that mitigates catastrophic forgetting of artificial and spiking neural networks with low computational cost. Science Advances. 2023;9(34). doi:10.1126/sciadv.adi2947

BibTeX
@article{Zhang2023,
  author = {Zhang, Tielin and Cheng, Xiang and Jia, Shuncheng and Li, Chengyu T and Poo, Mu-Ming and Xu, Bo},
  title = {A brain-inspired algorithm that mitigates catastrophic forgetting of artificial and spiking neural networks with low computational cost},
  journal = {Science Advances},
  year = {2023},
  volume = {9},
  number = {34},
  doi = {10.1126/sciadv.adi2947}
}
Kostenlos starten →