Window-Based Feature Representation and Ablation Analysis for EEG-Based Stress Classification

DOI

10.2991/978-94-6239-828-3_18How to use a DOI?

Keywords

EEG; Stress Classification; Windowing; Ablation Analysis

Abstract

Stress classification using electroencephalogram (EEG) signals has become an important research direction because EEG can represent changes in brain activity related to psychophysiological conditions. However, EEG signals are non-stationary, and analysis using long observation segments can reduce the ability of the model to capture local temporal variations. To improve the feature representation for stress classification, the SAM40 dataset was used for this study. A 7.5-second temporal resolution was implemented to better capture the data than non-segmented observations. A comparison of two feature sets, Power Spectral Density (PSD only) with its multi-domain combined formulation (PSD, Hjorth parameters, and Higuchi Fractal Dimension) was carried out. Ablation analysis assessed the incorporation of Synthetic Minority Over-sampling Technique (SMOTE), Particle Swarm Optimization (PSO), and stacking ensemble learning. It is discovered that window-based segmentation achieves better performance than non-windowed baselines. An optimal configuration was made using 7.5-second windowing with combined features together with an XGBoost classifier to achieve an accuracy of 0.83. The reduction in feature dimensionality produced by PSO from 288 to 157 was effective, but it was found to be model-dependent on predictive performance. This study’s main finding is that a 7.5-second windowing strategy has a strong impact on feature extraction, but optimization for increasing pipeline complexity does not automatically achieve better accuracy.0

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Fi Imanur Sifaunnufus Ms
AU  - Fitra Abdurrachman Bachtiar
PY  - 2026
DA  - 2026/11/04
TI  - Window-Based Feature Representation and Ablation Analysis for EEG-Based Stress Classification
BT  - Proceedings of the 2026 11th International Conference on Sustainable Information Engineering and Technology (SIET 2026)
PB  - Atlantis Press
SP  - 199
EP  - 209
SN  - 2589-4900
UR  - https://doi.org/10.2991/978-94-6239-828-3_18
DO  - 10.2991/978-94-6239-828-3_18
ID  - Ms2026
ER  -