Toward utilitarian online learning with class evolution

Zhi, Cao

Toward utilitarian online learning with class evolution

Zhi, Cao

Permalink

Publication Type:: Thesis
Issue Date:: 2025

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download thesisAdobe PDF (1.27 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Zhi, Cao
dc.date.accessioned	2026-01-15T03:42:36Z
dc.date.available	2026-01-15T03:42:36Z
dc.date.issued	2025
dc.identifier.uri	http://hdl.handle.net/10453/191855
dc.description	University of Technology Sydney. Faculty of Engineering and Information Technology.	en_US.UTF-8
dc.description.abstract	In real-world data stream mining, the composition of classes undergoes unpredictable changes, giving rise to the challenge of class evolution, which encompasses class emergence, class disappearance, and class reoccurrence. While most existing approaches necessitate the storage of past data to adapt their model to class evolution, several studies have focused on developing online learning algorithms that process instances in the data stream one-by-one and do not require the storage of historical data for model adaptation. However, existing online learning algorithms on handling data streams with class evolution are developed under potentially restrictive assumptions regarding 1) the prior distribution of classes, 2) the availability of labels for instances in the data stream for model adaptation, or 3) the assumption that the feature space remains fixed rather than varying. The non-stationary and unpredictable nature of the real-world data stream leads to unsatisfactory performance of these algorithms. Therefore, this research addresses these limitations by proposing online learning algorithms designed to manage data streams with class evolution for utilitarian, real-world applications.	en_US.UTF-8
dc.format	Thesis (PhD)
dc.language.iso	en_US	en_US.UTF-8
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/191855/1/thesis.pdf
dc.rights	info:eu-repo/semantics/openAccess
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	© 2025 Cao Zhi
dc.rights	au.edu.uts.lib/cph
dc.title	Toward utilitarian online learning with class evolution	en_US.UTF-8
dc.type	Thesis
utslib.copyright.status	open_access	*

Abstract:

In real-world data stream mining, the composition of classes undergoes unpredictable changes, giving rise to the challenge of class evolution, which encompasses class emergence, class disappearance, and class reoccurrence. While most existing approaches necessitate the storage of past data to adapt their model to class evolution, several studies have focused on developing online learning algorithms that process instances in the data stream one-by-one and do not require the storage of historical data for model adaptation. However, existing online learning algorithms on handling data streams with class evolution are developed under potentially restrictive assumptions regarding 1) the prior distribution of classes, 2) the availability of labels for instances in the data stream for model adaptation, or 3) the assumption that the feature space remains fixed rather than varying. The non-stationary and unpredictable nature of the real-world data stream leads to unsatisfactory performance of these algorithms. Therefore, this research addresses these limitations by proposing online learning algorithms designed to manage data streams with class evolution for utilitarian, real-world applications.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/191855