Reducing PEG Values by Comparing and Combining Various Approaches

DOI

10.2991/assehr.k.220401.173How to use a DOI?

Keywords

PEG values; database; regularization methods; best parameters; best combination

Abstract

This research aims to compare different approaches and combine the best approaches to reduce the Probability of Generalization Error (PEG) values of spoken classification. The size of training set is expanded by doubling specific rows for the training database to compare regularization and choose the best one for use and the Bagging method, Kfold method and Restarts method are compared to find the best one, which is combined with the regularization through the new database. The parameters are changed to combine the regularization with the method in accordance with the PEG values, and the best parameters are chosen which are also the best combination of the regularization with the method. Therefore, the best model of the best combination is picked out. Finally, the lowest PEG value is produced by using committee vote.

Copyright

© 2022 The Authors. Published by Atlantis Press SARL.

Open Access

This is an open access article distributed under the CC BY-NC 4.0 license.

Download article (PDF)

TY  - CONF
AU  - Wenjia Qi
PY  - 2022
DA  - 2022/04/08
TI  - Reducing PEG Values by Comparing and Combining Various Approaches
BT  - Proceedings of the 2022 International Conference on Social Sciences and Humanities and Arts (SSHA 2022)
PB  - Atlantis Press
SP  - 899
EP  - 904
SN  - 2352-5398
UR  - https://doi.org/10.2991/assehr.k.220401.173
DO  - 10.2991/assehr.k.220401.173
ID  - Qi2022
ER  -