EUVIMEDEUVIMED entdecken
Crossref

Exploring temporal information dynamics in Spiking Neural Networks: Fast Temporal Efficient Training

Changjiang Han, Li-Juan Liu, Hamid Reza Karimi

Journal of Neuroscience Methods · 2025

DOI: 10.1016/j.jneumeth.2025.110401 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Neuroscience Methods, 2025. Verfasst von Changjiang Han und 2 weiteren Autor:innen. Beteiligte Einrichtung: School of Railway Intelligent Engineering, Dalian Jiaotong University, Dalian, 116000, Liaoning, China. Hauptthemen laut MEDLINE: Neural Networks, Computer, Action Potentials, Neurons, Models, Neurological. Die Arbeit nennt 42 Literaturangaben.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

BACKGROUND: Spiking Neural Networks (SNNs) hold significant potential in brain simulation and temporal data processing. While recent research has focused on developing neuron models and leveraging temporal dynamics to enhance performance, there is a lack of explicit studies on neuromorphic datasets. This research aims to address this question by exploring temporal information dynamics in SNNs. NEW METHOD: To quantify the dynamics of temporal information during training, this study measures the Fisher information in SNNs trained on neuromorphic datasets. The information centroid is calculated to analyze the influence of key factors, such as the parameter k, on temporal information dynamics. RESULTS: Experimental results reveal that the information centroid exhibits two distinct behaviors: stability and fluctuation. This study terms this phenomenon the Stable Information Centroid (SIC), which is closely related to the parameter k. Based on these findings, we propose the Fast Temporal Efficient Training (FTET) algorithm. COMPARISON WITH EXISTING METHODS: Firstly, the method proposed in this paper does not require the introduction of additional complex training techniques. Secondly, it can reduce the computational load by 30% in the final 50 epochs. However, the drawback is the issue of slow convergence during the early stages of training. CONCLUSION: This study reveals that the learning processes of SNNs vary across different datasets, providing new insights into the mechanisms of human brain learning. A limitation is the restricted sample size, focusing only on a few datasets and image classification tasks. The code is available at https://github.com/gtii123/fast-temporal-efficient-training.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Neural Networks, ComputerAction PotentialsNeuronsModels, NeurologicalHumansAlgorithmsBrainTime Factors

Stichwörter der Autor:innen

Brain simulationBrain-inspired computingSignal processingSpiking neural networksVision

Autor:innen und Institutionen

  • School of Railway Intelligent Engineering, Dalian Jiaotong University, Dalian, 116000, Liaoning, China.
  • School of Railway Intelligent Engineering, Dalian Jiaotong University, Dalian, 116000, Liaoning, China. Electronic address: liulj@djtu.edu.cn.
  • Department of Mechanical Engineering, Politecnico di Milano, Milan, 20121, Lombardia, Italy.

Zitieren

APA

Han, C., Liu, L. J., & Karimi, H. R. (2025). Exploring temporal information dynamics in Spiking Neural Networks: Fast Temporal Efficient Training. Journal of Neuroscience Methods. https://doi.org/10.1016/j.jneumeth.2025.110401

Vancouver

Han C, Liu LJ, Karimi HR. Exploring temporal information dynamics in Spiking Neural Networks: Fast Temporal Efficient Training. Journal of Neuroscience Methods. 2025. doi:10.1016/j.jneumeth.2025.110401

BibTeX
@article{Han2025,
  author = {Han, Changjiang and Liu, Li-Juan and Karimi, Hamid Reza},
  title = {Exploring temporal information dynamics in Spiking Neural Networks: Fast Temporal Efficient Training},
  journal = {Journal of Neuroscience Methods},
  year = {2025},
  doi = {10.1016/j.jneumeth.2025.110401}
}
Kostenlos starten →