TechRxiv
RRR_submit_Final.pdf (16.42 MB)
Download file

Resampling, Relabeling, & Raking Algorithm to One-Class Classification

Download (16.42 MB)
preprint
posted on 27.04.2022, 04:02 by Hae-Hwan Lee, Seunghwan Park, Jongho ImJongho Im
The performance of a classification model significantly depends on the degree to which the support of each data class overlaps. Successfully distinguishing classes is difficult if the support is similar while classes differ. In the one-class classification (OCC) problem, wherein the data comprise only a single class, classifier performance is significantly degraded if the population support of each class is similar. In this study, we propose a preprocessing algorithm that enhances classifier performance by utilizing macro information that is most easily obtainable in these two problem situations. The algorithm aims to improve classifier performance by reprocessing the given data into data with mitigated class imbalance through raking and sampling techniques. This improvement in performance is demonstrated by comparing representative classifiers used in the existing OCC problem with traditional binary classifier models, unavailable on the single-class dataset.

Funding

National Research Foundation of Korea (NRF-2021R1C1C1014407)

National Research Foundation of Korea (NRF-NRF-2019R1G1A1002232)

History

Email Address of Submitting Author

ijh38@yonsei.ac.kr

ORCID of Submitting Author

0000-0001-8362-4756

Submitting Author's Institution

Yonsei Unversity

Submitting Author's Country

Korea

Usage metrics

Licence

Exports