Hybrid CNN-LSTM-Attention Model Achieves 78.5% Accuracy in Single-Channel EEG Cognitive Load Assessment
July 3, 2026
A hybrid CNN, LSTM, and Attention architecture achieves 78.5% accuracy in classifying cognitive load using single-channel NeuroSky MindWave Mobile 2 data. The model outperforms conventional feature-based classifiers, which only reached 55% accuracy, by combining raw waveforms with band-power features in a within-subject setting.
HOW THIS AFFECTS YOU
●
builderYou can explore integrating low-cost, single-channel EEG hardware into educational monitoring tools.
●
researcherYou can leverage the hybrid architecture to improve classification performance on noisy, consumer-grade EEG signals.