Stochastic computing & stochastic resonance in digital & analog neuromorphic systems.

Abstract

Stochastic computing is an alternative method to standard deterministic methods. This thesis explores stochastic resonance in a spiking neuron, and leverages the effects it has to implement stochastic computing in an analog spiking neuron, and simple neuron network. In addition, a method for interfacing a Stochastic Artificial Neural Network implemented on a Field Programmable Gate Array with a computer running MATLAB is presented.