Modeling and Simulation of Spiking Neural Network Models using SpiNNaker

Abstract

Simulating Spiking Neural Networks (SNN) models is a research field attracting the interest of researchers from various fields, from biology to computer science. The final objective is understanding the mechanisms defining the human brain working. Multiple neural models have been proposed, each with their peculiarities, from the very complex and biologically realistic Hodgkin-Huxley neuron model to the very simple leaky integrate-and-fire neuron. Researchers can, depending on the objective, choose which model to use in their simulation. For an efficient simulation of large population of neurons using these models, there need to be a real parallel system architecture and biologically realistic simulator. This research work revolves around using a universally accepted biologically accurate NEST (software simulator) and SpiNNaker (hardware based simulator). During this research, Hodgkin Huxley model has been implemented for the first time on SpiNNaker using fixed point notation and its results have been verified with those from NEST. Similarly, a newly proposed AJ neural model has been implemented for the first time over NEST and SpiNNaker and we successfully verified its results with those from MATLAB. The research contributed in devising implementation libraries for these two models for researchers interested to simulate neural populations over SpiNNaker and NEST using these models.