Probabilistic Value Selection for Space Efficient Model

Gunarto Sindoro Njoo, Baihua Zheng, Kuo Wei Hsu, Wen Chih Peng

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

An alternative to current mainstream preprocessing methods is proposed: Value Selection (VS). Unlike the existing methods such as feature selection that removes features and instance selection that eliminates instances, value selection eliminates the values (with respect to each feature) in the dataset with two purposes: reducing the model size and preserving its accuracy. Two probabilistic methods based on information theory's metric are proposed: PVS and \mathrm {P}^{+}VS. Extensive experiments on the benchmark datasets with various sizes are elaborated. Those results are compared with the existing preprocessing methods such as feature selection, feature transformation, and instance selection methods. Experiment results show that value selection can achieve the balance between accuracy and model size reduction.

Original languageEnglish
Title of host publicationProceedings - 2020 21st IEEE International Conference on Mobile Data Management, MDM 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages148-157
Number of pages10
ISBN (Electronic)9781728146638
DOIs
StatePublished - Jun 2020
Event21st IEEE International Conference on Mobile Data Management, MDM 2020 - Versailles, France
Duration: 30 Jun 20203 Jul 2020

Publication series

NameProceedings - IEEE International Conference on Mobile Data Management
Volume2020-June
ISSN (Print)1551-6245

Conference

Conference21st IEEE International Conference on Mobile Data Management, MDM 2020
CountryFrance
CityVersailles
Period30/06/203/07/20

Keywords

  • Data mining
  • Entropy
  • Information theory
  • Model size reduction
  • Preprocessing
  • Value selection

Fingerprint Dive into the research topics of 'Probabilistic Value Selection for Space Efficient Model'. Together they form a unique fingerprint.

Cite this