SAB: Self-Adaptive Bias

Curriculum learning is a method of prioritizing learning data to improve learning performance. In this paper, we propose a new algorithm that determines how to select learning data and when to start and stop curriculum learning by considering learning errors. We use entropy to select data samples wi...

Full description

Saved in:
Bibliographic Details
Main Authors: Suchan Choi, Jinyoung Oh, Jeong-Won Cha
Format: Article
Language:English
Published: MDPI AG 2024-12-01
Series:AI
Subjects:
Online Access:https://www.mdpi.com/2673-2688/5/4/133
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850050658384740352
author Suchan Choi
Jinyoung Oh
Jeong-Won Cha
author_facet Suchan Choi
Jinyoung Oh
Jeong-Won Cha
author_sort Suchan Choi
collection DOAJ
description Curriculum learning is a method of prioritizing learning data to improve learning performance. In this paper, we propose a new algorithm that determines how to select learning data and when to start and stop curriculum learning by considering learning errors. We use entropy to select data samples with less consistent predictions and automatically determine the warming-up period based on the characteristics of the data. Additionally, to mitigate learning bias, we introduced a variable that adjusts the range of sample selection according to the progress of the training. To validate our method, we conducted extensive experiments on both balanced and imbalanced data classification tasks, and our proposed approach showed an average improvement of about 1.8%, with a maximum improvement of up to 4.4%, compared to previously suggested methods.
format Article
id doaj-art-c82171f604d54ee7a26e2db48d152a17
institution DOAJ
issn 2673-2688
language English
publishDate 2024-12-01
publisher MDPI AG
record_format Article
series AI
spelling doaj-art-c82171f604d54ee7a26e2db48d152a172025-08-20T02:53:22ZengMDPI AGAI2673-26882024-12-01542761277210.3390/ai5040133SAB: Self-Adaptive BiasSuchan Choi0Jinyoung Oh1Jeong-Won Cha2Department of Computer Engineering, Changwon National University, Changwon 51140, Republic of KoreaDepartment of Computer Engineering, Changwon National University, Changwon 51140, Republic of KoreaDepartment of Computer Engineering, Changwon National University, Changwon 51140, Republic of KoreaCurriculum learning is a method of prioritizing learning data to improve learning performance. In this paper, we propose a new algorithm that determines how to select learning data and when to start and stop curriculum learning by considering learning errors. We use entropy to select data samples with less consistent predictions and automatically determine the warming-up period based on the characteristics of the data. Additionally, to mitigate learning bias, we introduced a variable that adjusts the range of sample selection according to the progress of the training. To validate our method, we conducted extensive experiments on both balanced and imbalanced data classification tasks, and our proposed approach showed an average improvement of about 1.8%, with a maximum improvement of up to 4.4%, compared to previously suggested methods.https://www.mdpi.com/2673-2688/5/4/133curriculum learningadaptive batch selectionpre-trained models
spellingShingle Suchan Choi
Jinyoung Oh
Jeong-Won Cha
SAB: Self-Adaptive Bias
AI
curriculum learning
adaptive batch selection
pre-trained models
title SAB: Self-Adaptive Bias
title_full SAB: Self-Adaptive Bias
title_fullStr SAB: Self-Adaptive Bias
title_full_unstemmed SAB: Self-Adaptive Bias
title_short SAB: Self-Adaptive Bias
title_sort sab self adaptive bias
topic curriculum learning
adaptive batch selection
pre-trained models
url https://www.mdpi.com/2673-2688/5/4/133
work_keys_str_mv AT suchanchoi sabselfadaptivebias
AT jinyoungoh sabselfadaptivebias
AT jeongwoncha sabselfadaptivebias