Window-Based Feature Engineering for Cognitive Workload Detection

TL;DR

This paper addresses cognitive workload detection from eye-tracking data by introducing a window-based temporal feature engineering framework applied to the COLET dataset. It systematically compares traditional ML models with tabular DL architectures, notably TabNet and SNN, and demonstrates that deep learning approaches better capture temporal dynamics, achieving higher accuracy and F1-scores. The key contributions include a novel windowed feature set, a comprehensive cross-method evaluation, and evidence that TabNet and related DL models support real-time CW monitoring in dynamic tasks. The findings have practical implications for real-time workload adaptation in safety-critical contexts such as driving and complex operational environments.

Abstract

Cognitive workload is a topic of increasing interest across various fields such as health, psychology, and defense applications. In this research, we focus on classifying cognitive workload using the COLET dataset, employing a window-based approach for feature generation and machine/deep learning techniques for classification. We apply window-based temporal partitioning to enhance features used in existing research, followed by machine learning and deep learning models to classify different levels of cognitive workload. The results demonstrate that deep learning models, particularly tabular architectures, outperformed traditional machine learning methods in precision, F1-score, accuracy, and classification precision. This study highlights the effectiveness of window-based temporal feature extraction and the potential of deep learning techniques for real-time cognitive workload assessment in complex and dynamic tasks.

Figures (3)

• Figure 1: Overview of the proposed method
• Figure 2: Feature extraction from COLET Data
• Figure 3: Window Generation