Integrating Non-Spiking Interneurons in Spiking Neural Networks

Abstract

Researchers working with neural networks have historically focused on either non-spiking neurons tractable for running on computers or more biologically plausible spiking neurons requiring special hardware. However, in nature homogeneous networks of neurons do not exist. Instead, spiking and non-spiking neurons cooperate, each bringing a different set of advantages. A well researched biological example of such a mixed network is the sensorimotor pathway, responsible for mapping sensory inputs to behavioral changes. This pathway is also well researched in robotics where it is applied to achieve closed-loop operation of legged robots by adapting amplitude, frequency, and phase of the motor output. In this paper we investigate how spiking and non-spiking neurons can be combined to create a sensorimotor neuron pathway capable of shaping network output based on analog input. We propose sub-threshold operation of an existing spiking neuron model to create a non-spiking neuron able to interpret analog information and communicate with spiking neurons. The validity of this methodology is confirmed through a simulation of a closed-loop amplitude regulating network. Although it is currently not possible to run non-spiking neurons on most neuromorphic hardware platforms, we expect that mixed networks can be achieved through combining specialized hardware with conventional tools. The ability to work with mixed networks provides an opportunity for researchers to investigate new network architectures for adaptive controllers, potentially improving locomotion strategies of legged robots.

CloudBrain: Online neural computation in the cloud

“…Integrating non-spiking neurons in SNNs can be a biologically plausible way to interface analogue sensors and can provide more control over network behaviour (Strohmer et al, 2020).…”

“…For example biological neurons can be non-spiking (Sterling and Laughlin, 2015) and there could be advantages of combining spiking and non-spiking models (Woźniak et al, 2020). Integrating non-spiking neurons in SNNs can be a biologically plausible way to interface analogue sensors and can provide more control over network behaviour (Strohmer et al, 2020).…”