Online ensemble learning of data streams with gradually evolved classes
- Submitting institution
-
The University of Birmingham
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 54722265
- Type
- D - Journal article
- DOI
-
10.1109/TKDE.2016.2526675
- Title of journal
- IEEE Transactions on Knowledge and Data Engineering
- Article number
- -
- First page
- 1532
- Volume
- 28
- Issue
- 6
- ISSN
- 1041-4347
- Open access status
- Out of scope for open access requirements
- Month of publication
- February
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 51
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Class evolution is a challenge in data stream learning, where new classes emerge or existing classes disappear. This paper recognized the research gap and studied gradual class evolution. A class-based ensemble approach was proposed, which can rapidly adjust to class evolution by maintaining a base learner for each class. The empirical studies demonstrate its effectiveness in various class evolution scenarios in comparison to five existing methods. The result was validated through Wilcoxon sign rank test. This paper was published in a leading machine learning journal, and cited by world-leading groups in universities such as University of Porto and Rowan University
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -