EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Biologically inspired heterogeneous learning for accurate, efficient and low-latency neural network

Bo Wang, Yuxuan Zhang, Hongjue Li, Hongkun Dou, Yuchen Guo, Yue Deng

National Science Review · 30.12.2024 · 12 (1)

DOI: 10.1093/nsr/nwae301 ↗

Kurz gefasst

Fachartikel in der Zeitschrift National Science Review, online erschienen am 30.08.2024. Verfasst von Bo Wang und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Astronautics, Beihang University, Beijing 100191, China. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 67 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The pursuit of artificial neural networks that mirror the accuracy, efficiency and low latency of biological neural networks remains a cornerstone of artificial intelligence (AI) research. Here, we incorporated recent neuroscientific findings of self-inhibiting autapse and neuron heterogeneity for innovating a spiking neural network (SNN) with enhanced learning and memorizing capacities. A bi-level programming paradigm was formulated to respectively learn neuron-level biophysical variables and network-level synapse weights for nested heterogeneous learning. We successfully demonstrated that our biologically inspired neuron model could reproduce neural statistics at both individual and group levels, contributing to the effective decoding of brain–computer interface data. Furthermore, the heterogeneous SNN showed higher accuracy (1%–10% improvement), superior efficiency (maximal 17.83-fold reduction in energy) and lower latency (maximal 5-fold improvement) in performing several AI tasks. For the first time, we benchmarked SNN for conducting cell type identification from scRNA-seq data. The proposed model correctly identified very rare cell types associated with severe brain diseases where typical SNNs failed.

Schlagwörter

Stichwörter der Autor:innen

biologically inspiredheterogeneous learningmemoryscRNA-seqspiking neural network

Autor:innen und Institutionen

  • School of Astronautics, Beihang University, Beijing 100191, China.
  • Institute for Brain and Cognitive Sciences, BNRist, Tsinghua University, Beijing 100084, China.
  • School of Artificial Intelligence, Beihang University, Beijing 100191, China.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Wang, B., Zhang, Y., Li, H., Dou, H., Guo, Y., & Deng, Y. (2024). Biologically inspired heterogeneous learning for accurate, efficient and low-latency neural network. National Science Review, 12(1). https://doi.org/10.1093/nsr/nwae301

Vancouver

Wang B, Zhang Y, Li H, Dou H, Guo Y, Deng Y. Biologically inspired heterogeneous learning for accurate, efficient and low-latency neural network. National Science Review. 2024;12(1). doi:10.1093/nsr/nwae301

BibTeX
@article{Wang2024,
  author = {Wang, Bo and Zhang, Yuxuan and Li, Hongjue and Dou, Hongkun and Guo, Yuchen and Deng, Yue},
  title = {Biologically inspired heterogeneous learning for accurate, efficient and low-latency neural network},
  journal = {National Science Review},
  year = {2024},
  volume = {12},
  number = {1},
  doi = {10.1093/nsr/nwae301}
}
Kostenlos starten →