Granular computing based approach for classification towards reduction of bias in ensemble learning
- Submitting institution
-
University of Portsmouth
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 14316763
- Type
- D - Journal article
- DOI
-
10.1007/s41066-016-0034-1
- Title of journal
- Granular Computing
- Article number
- -
- First page
- 131
- Volume
- 2
- Issue
- 3
- ISSN
- 2364-4966
- Open access status
- Compliant
- Month of publication
- November
- 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
-
1
- Research group(s)
-
B - Computational Intelligence
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Ensemble learning, i.e. learning using several classifiers, requires the aggregation of the outputs of individual classifiers to deliver one output. The current ways of doing this aggregation (majority and weighted voting) are biased towards the class with the highest frequency or weight, resulting in overfitting. The new aggregation does not only reduce the bias, but also leads to improvements in classification performance, as acknowledged in in Liu&Fan, Applied Intelligence, 2020 (DOI 10.1007/s10489-020-01911-0) and Li&Xu, Library High Tech, 2019 (DOI 10.1108/LHT-11-2018-0166).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -