Morphological learning in spiking neurons: a new hardware efficient maching learning method

Abstract

The brain has fascinated mankind from time immemorial due to it computational prowess and complexity. The latest developments in the research of spiking neural network models have shown that unlike the classic neural network models, these models communicate via precisely timed neuron spikes, thus making them a closer representation of the biological neurons. ‘Morphological Learning in Spiking Neurons: A New Hardware Efficient Machine Learning Method’ explores the greater performance of spiking neurons with lumped non-linearity than their counterparts with linear synaptic summation of signals. The better performance is due to the additional degree of freedom in such neurons. The algorithm presented is in this project is hardware friendly for learning. MATLAB software developed by MathWorks has been used as a computational tool to simulate the different neuron models and WTA networks as it offers an environment to generate graphical results easily.