Online Ensemble of Ensemble OVA Framework for Class Evolution with Dominant Emerging Classes

Cao, Z; Zhang, S; Lin, C-T

Online Ensemble of Ensemble OVA Framework for Class Evolution with Dominant Emerging Classes

Cao, Z Zhang, S Lin, C-T

Permalink

Publisher:: IEEE
Publication Type:: Conference Proceeding
Citation:: 23RD IEEE INTERNATIONAL CONFERENCE ON DATA MINING, ICDM 2023, 2024, pp. 968-973
Issue Date:: 2024

In Progress

	Filename	Description	Size
	Online_Ensemble_of_Ensemble_OVA_Framework_for_Class_Evolution_with_Dominant_Eme.pdf	Published version	409 kB		View/Open

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is being processed and is not currently available.

Full metadata record

Field	Value	Language
dc.contributor.author	Cao, Z
dc.contributor.author	Zhang, S
dc.contributor.author	Lin, C-T
dc.contributor.editor	Chen, G
dc.contributor.editor	Khan, L
dc.contributor.editor	Gao, X
dc.contributor.editor	Qiu, M
dc.contributor.editor	Pedrycz, W
dc.contributor.editor	Wu, X
dc.date	2023-12-01
dc.date.accessioned	2025-01-16T00:37:43Z
dc.date.available	2025-01-16T00:37:43Z
dc.date.issued	2024
dc.identifier.citation	23RD IEEE INTERNATIONAL CONFERENCE ON DATA MINING, ICDM 2023, 2024, pp. 968-973
dc.identifier.issn	1550-4786
dc.identifier.uri	http://hdl.handle.net/10453/183711
dc.description.abstract	In real-world data stream mining, the composition of classes undergoes unpredictable changes, giving rise to the challenge of class evolution, encompassing class emergence, disappearance, and reoccurrence. However, most existing approaches require the storage of past data to adapt their model. While some studies have focused on online learning approaches, they are built on an underlying assumption that the number of instances in any single class is consistently less than the sum of other classes. This assumption becomes invalid when a class emerges with a dominant amount, e.g., news about a pandemic outbreak, harming the performance of existing methods. In this paper, we thoroughly investigate this scenario and propose a novel online ensemble of ensemble one-versus-all framework (EEOF) to handle class evolution adaptively. The novel ensemble of ensemble architecture boosts diversity in each one-versus-all classifier. A novel adaptive model adaptation method is also designed to balance the error feedback between the emerging class and the other classes. A confidence-triggered fallback mode is integrated to prevent performance drop due to a wrong decision regarding class disappearance. Experimental studies are conducted on both synthetic and real-world data streams to show that our method achieves higher accuracy in diverse class evolution scenarios compared with the state-of-the-art method, particularly when classes emerge with dominant amounts.
dc.language	en
dc.publisher	IEEE
dc.relation.ispartof	23RD IEEE INTERNATIONAL CONFERENCE ON DATA MINING, ICDM 2023
dc.relation.ispartof	IEEE International Conference on Data Mining
dc.relation.ispartofseries	IEEE International Conference on Data Mining
dc.relation.isbasedon	10.1109/1CDM58522.2023.00109
dc.rights	info:eu-repo/semantics/restrictedAccess
dc.title	Online Ensemble of Ensemble OVA Framework for Class Evolution with Dominant Emerging Classes
dc.type	Conference Proceeding
utslib.location.activity	Shanghai
pubs.organisational-group	University of Technology Sydney
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology/School of Computer Science
pubs.organisational-group	University of Technology Sydney/UTS Groups
pubs.organisational-group	University of Technology Sydney/UTS Groups/Australian Artificial Intelligence Institute (AAII)
utslib.copyright.status	in_progress	*
pubs.consider-herdc	false
dc.date.updated	2025-01-16T00:37:42Z
pubs.finish-date	2023-12-04
pubs.place-of-publication	Piscataway, USA
pubs.publication-status	Published
pubs.start-date	2023-12-01
dc.location	Piscataway, USA

Abstract:

In real-world data stream mining, the composition of classes undergoes unpredictable changes, giving rise to the challenge of class evolution, encompassing class emergence, disappearance, and reoccurrence. However, most existing approaches require the storage of past data to adapt their model. While some studies have focused on online learning approaches, they are built on an underlying assumption that the number of instances in any single class is consistently less than the sum of other classes. This assumption becomes invalid when a class emerges with a dominant amount, e.g., news about a pandemic outbreak, harming the performance of existing methods. In this paper, we thoroughly investigate this scenario and propose a novel online ensemble of ensemble one-versus-all framework (EEOF) to handle class evolution adaptively. The novel ensemble of ensemble architecture boosts diversity in each one-versus-all classifier. A novel adaptive model adaptation method is also designed to balance the error feedback between the emerging class and the other classes. A confidence-triggered fallback mode is integrated to prevent performance drop due to a wrong decision regarding class disappearance. Experimental studies are conducted on both synthetic and real-world data streams to show that our method achieves higher accuracy in diverse class evolution scenarios compared with the state-of-the-art method, particularly when classes emerge with dominant amounts.

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

http://hdl.handle.net/10453/183711