Bayesian Augmentation of Deep Learning to Improve Video Classification

Date
2022-01-04
Authors
Swize, Emmie
Champagne, Lance
Cox, Bruce
Bihl, Trevor
Contributor
Advisor
Department
Instructor
Depositor
Speaker
Researcher
Consultant
Interviewer
Annotator
Journal Title
Journal ISSN
Volume Title
Publisher
Volume
Number/Issue
Starting Page
Ending Page
Alternative Title
Abstract
Traditional automated video classification methods lack measures of uncertainty, meaning the network is unable to identify those cases in which its predictions are made with significant uncertainty. This leads to misclassification, as the traditional network classifies each observation with same amount of certainty, no matter what the observation is. Bayesian neural networks are a remedy to this issue by leveraging Bayesian inference to construct uncertainty measures for each prediction. Because exact Bayesian inference is typically intractable due to the large number of parameters in a neural network, Bayesian inference is approximated by utilizing dropout in a convolutional neural network. This research compared a traditional video classification neural network to its Bayesian equivalent based on performance and capabilities. The Bayesian network achieves higher accuracy than a comparable non-Bayesian video network and it further provides uncertainty measures for each classification.
Description
Keywords
Soft Computing: Theory Innovations and Problem Solving Benefits, bayesian, classification, deep learning, video, video surveillance
Citation
Extent
10 pages
Format
Geographic Location
Time Period
Related To
Proceedings of the 55th Hawaii International Conference on System Sciences
Table of Contents
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International
Rights Holder
Local Contexts
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.