Implementation of artificial intelligence cognitive neuroscience neuron cell using adaptive velocity threshold particle swarm optimization (AVT-PSO) on FPGA

Abstract

This paper presents the hardware development and implementation of artificial intelligence based neuron cell using swarm intelligence based algorithm (AVT-PSO) where the functionality of the architecture is tested by implementing four bit addition based on cognitive science neural network employing five neuron cell (sort of a model emulating function of a neuron cell in the brain) trained using an adaptive velocity threshold particle swarm optimization on Spartan-3e XC3S100E Field Programmable Gate Arrays (FPGA). Each neuron cell represents a processing element which is trained using swarm intelligence. Adaptive velocity threshold PSO algorithm is used in evolving threshold values to train the weights of neural cells. Implemented system is flexible in design, allowing the possibility to add or remove neurons to generate new network architectures.