Publication Date

Spring 2026

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)

Department

Computer Science

First Advisor

William Andreopoulos

Second Advisor

Sayma Akther

Third Advisor

Pratyush Kshirsagar

Keywords

Neuromorphic computing, spiking neural network, surrogate gra- dient learning, operating system, energy efficiency

Abstract

Many existing AI models rely on computationally expensive architectures that are poorly suited for real-time and always-on applications. This project investigates the use of neuromorphic computing as a more energy efficient alternative. In this project there are three components to view in order to test the effectiveness of using a neuromorphic computing approach. The first component is creating a spiking neural network and a deep neural network model. A comparison will be done between the models to view the difference in energy and computational costs. The second component is implementing two different types of GPUs during the training process of the spiking neural network. The reason for doing this is to see which one will be used for the comparison between the spiking neural network and deep neural network. The third component is creating an operating system that can be integrated with a spiking neural network.

Download

Available for download on Saturday, May 22, 2027

Share

COinS