ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Latent-Variable Classifiers Based on Total Correlation for Dynamic Environments
Cited 0 time in scopus Share share facebook twitter linkedin kakaostory
Authors
Cheol Ho Kim, Jung-Hoon Lee, Byounghwa Lee, Ock Kee Baek
Issue Date
2023-10
Citation
International Conference on Information and Communication Technology Convergence (ICTC) 2023, pp.103-105
Publisher
IEEE
Language
English
Type
Conference Paper
DOI
https://dx.doi.org/10.1109/ICTC58733.2023.10393699
Abstract
This paper introduces a novel approach to latentvariable classifiers that employs the correlation explanation technique, an information-theoretic latent-variable clustering based on total correlation. The proposed classifier improves batch-mode classification accuracy in comparison to traditional training approaches. Additionally, it demonstrates the ability to incrementally learn from additional datasets involving new features, without accessing or reprocessing past data. These findings underscore the method's potential to serve as a robust machine learning model suitable for dynamic real-world environments.
KSP Keywords
Batch-mode, Dynamic Environment, Learning model, Novel approach, Real-world, Total correlation, Variable clustering, classification accuracy, clustering based, information-theoretic, machine Learning