Theses and Dissertations

Advisor

Jones, Adam

Committee Member

Swan, Edward

Committee Member

Collins, Eric

Committee Member

Jankun-Kelley, T.J.

Committee Member

Ford, David

Date of Degree

5-15-2026

Original embargo terms

Immediate Worldwide Access

Document Type

Graduate Thesis - Open Access

Major

Computer Science (Research Computer Science)

Degree Name

Master of Science (M.S.)

College

James Worth Bagley College of Engineering

Department

Department of Computer Science and Engineering

Abstract

To alleviate the computational restraints inherent in 3D Deep Learning tasks, we propose a data compression method based on existing voxelization techniques. We analyze this technique through tests of structural similarity, speed, and machine learning application. In order to appropriately test its performance for machine learning tasks, we also propose an efficient method of training Convolutional Neural Networks by utilizing binary bit data as an input for a multiclass classification problem. Our current findings show that these two contributions have potential to aid the state of the art in handling large data in 3D AI/ML tasks, both in the space the data takes on disc and in the resources required to handle large data during training.

Download

Share

COinS