Year: 2025
Type: article
Abstract: Plasticity is key to the trainability of neural networks and has long been a focus in the field of brain-inspired research. Currently, neuromorphic networks primarily achieve plasticity through synaptic and myelin structures. However, these two are often studied separately, limiting further enhancement of neuronal node plasticity. This paper proposes a neuron model that incorporates both synapses and myelin, designs the corresponding neuronal circuit, and introduces a method for quantifying its discharge characteristics. Through theoretical analysis, simulations, and physical experiments, we validate the effectiveness of this quantification method. Furthermore, we summarize the formation mechanisms of synaptic and myelin plasticity, clarify the differences in their respective plasticity effects, and use the quantification method to compute the response speed, power consumption, and spike firing frequency of neuronal circuits. We also analyze the impact of synaptic and myelin plasticity and their synergistic effects on these three factors. Results demonstrate that the plasticity of synapses and myelin, as well as their synergistic interaction, can significantly optimize the performance of neuron nodes: the response duration is reduced to 2.9% of its initial value, the energy consumption per spike decreases to 38.4%, and the spike firing frequency increases to 1982.6% of the baseline level. This synergy contributes to improving the computational efficiency and energy management capabilities of neuromorphic networks.
Institutions: Hunan University, University of Hertfordshire
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
Awards: G5258255866, G7750634368
OpenAlex ID: W4411269914