Comparative Analysis of Biological Spiking Neuron Models for Classification Task

Artificial neural networks (ANNs) have shown promising result in many applications, but if compared with biological neural networks (BNNs) it still lags behind in many ways. By exploiting biological plausible neurons, spiking neural networks(SNNs) works to fill the void between ANNs and BNNs. In the field of machine learning, the spiking neural network has gained significant attention due to its potential for achieving high-performance computing with low power consumption. In this study, a comparative analysis of different biological spiking neurons for a classification task of handwritten digits from MNIST dataset has been presented. Specifically, the performance of four different neuron models has been compared and found Leaky Integrate-and-Fire neuron is giving best results with 98.04% accuracy with only one hidden layer in the network. These findings provide valuable insights into the use of different biological spiking neurons for classification tasks and can aid in the development of more efficient spiking neural networks for various applications.