ระบบจัดเก็บและค้นหางานวิจัยอัจฉริยะ

AlphaGrad: Normalized Gradient Descent for Adaptive Multi-Loss Functions in EEG-Based Motor Imagery Classification

Rattanaphon Chaisaen | Brain-computer interfaces multi-task learning (MTL) adaptive loss blending

จำนวนผู้เข้าชม

AlphaGrad: Normalized Gradient Descent for Adaptive Multi-Loss Functions in EEG-Based Motor Imagery Classification

Collection	RMUTK Research Repository (RMUTK IR) วารสารวิชาการ — e-Journal Articles
ID RMUTK Digital	RMUTK000119
Title	AlphaGrad: Normalized Gradient Descent for Adaptive Multi-Loss Functions in EEG-Based Motor Imagery Classification
Alternative title	AlphaGrad: Normalized Gradient Descent for Adaptive Multi-Loss Functions in EEG-Based Motor Imagery Classification
Authors	อ.ดร.อภิวัฒน์ ดิษฐาพร Corresponding Rattanaphon Chaisaen ผู้แต่งหลัก Phairot Autthasan Theerawit Wilaiprasitporn Corresponding
Faculty	คณะบริหารธุรกิจ
Journal Title	IEEE Journal of Biomedical and Health Informatics
ISSN	2168-2208
Volume / Issue / Pages	Vol.29 \| No.10 \| pp.7116 - 7128
Published	2025-05-22
Year	2568
Level	ระดับนานาชาติ (SCOPUS)
Quartile	Q1
DOI	10.1109/JBHI.2025.3572197
Abstract	In this study, we propose AlphaGrad, a novel adaptive loss blending strategy for optimizing multi-task learning (MTL) models in motor imagery (MI)-based electroencephalography (EEG) classification. AlphaGrad is the first method to automatically adjust multi-loss functions with differing metric scales, including mean square error, cross-entropy, and deep metric learning, within the context of MI-EEG. We evaluate AlphaGrad using two state-of-the-art MTL-based neural networks, MIN2Net and FBMSNet, across four benchmark datasets. Experimental results show that AlphaGrad consistently outperforms existing strategies such as AdaMT, GradApprox, and fixed-weight baselines in classification accuracy and training stability. Compared to baseline static weighting, AlphaGrad achieves over 10% accuracy improvement on subject-independent MI tasks when evaluated on the largest benchmark dataset. Furthermore, AlphaGrad demonstrates robust adaptability across various EEG paradigms—including steady-state visually evoked potential (SSVEP) and event-related potential (ERP), making it broadly applicable to brain-computer interface (BCI) systems. We also provide gradient trajectory visualizations highlighting AlphaGrad's ability to maintain training stability and avoid local minima. These findings underscore AlphaGrad's promise as a general-purpose solution for adaptive multi-loss optimization in biomedical time-series learning.
Abstract (EN)	In this study, we propose AlphaGrad, a novel adaptive loss blending strategy for optimizing multi-task learning (MTL) models in motor imagery (MI)-based electroencephalography (EEG) classification. AlphaGrad is the first method to automatically adjust multi-loss functions with differing metric scales, including mean square error, cross-entropy, and deep metric learning, within the context of MI-EEG. We evaluate AlphaGrad using two state-of-the-art MTL-based neural networks, MIN2Net and FBMSNet, across four benchmark datasets. Experimental results show that AlphaGrad consistently outperforms existing strategies such as AdaMT, GradApprox, and fixed-weight baselines in classification accuracy and training stability. Compared to baseline static weighting, AlphaGrad achieves over 10% accuracy improvement on subject-independent MI tasks when evaluated on the largest benchmark dataset. Furthermore, AlphaGrad demonstrates robust adaptability across various EEG paradigms—including steady-state visually evoked potential (SSVEP) and event-related potential (ERP), making it broadly applicable to brain-computer interface (BCI) systems. We also provide gradient trajectory visualizations highlighting AlphaGrad's ability to maintain training stability and avoid local minima. These findings underscore AlphaGrad's promise as a general-purpose solution for adaptive multi-loss optimization in biomedical time-series learning.
Keywords (TH)	Brain-computer interfaces multi-task learning (MTL)adaptive loss blending motor imagery EEG normalized gradient descent
Keywords (EN)	Brain-computer interfaces multi-task learning (MTL)adaptive loss blending motor imagery EEG normalized gradient descent
Access Level	Open Access (เปิดสาธารณะ)

ลิงก์แหล่งเผยแพร่

เอกสารฉบับเต็ม (ลิงก์ภายนอก)

https://ieeexplore.ieee.org/document/11008918

เปิดลิงก์