Cognitive Load Estimation from Simulated EEG Using Band-Power Features and SVM Classification

This work presents a transparent, low-complexity framework for cognitive load classification using synthetic EEG signals. We simulate multichannel EEG data representing low and high cognitive load states, extract frequency-domain features using Welch\u2019s method, and train a support vector machine (SVM) classifier. The proposed model achieves 100% accuracy, F1-score, and ROC AUC under moderate noise (\u03c3\u202f=\u202f0.5), with inference latency under 2\u202fms. Compared to deep learning approaches, our pipeline offers interpretability, rapid prototyping, and computational efficiency. All code, data, and reproducibility assets are publicly released.