Joint Modality Features in Frequency Domain for Stress Detection

Rich feature extraction is essential to train a good machine learning (ML) framework. These features are generally extracted separately from each modality. We hypothesize that richer can be learned when modalities jointly explored. joint modality perform better than those individual modalities. study two modalities, physiological signals – Electrodermal activity (EDA) and electrocardiogram (ECG) investigate this hypothesis. our hypothesis achieve three objectives for subject-independent stress detection. For the first time in literature, we apply proposed framework frequency domain. The frequency-domain decomposition of signal effectively separates it into periodic aperiodic components.We correlate their behaviour by focusing on band spectrum. Second, show outperforms late fusion, early fusion other notable works field. Finally, validate approach four benchmark datasets its generalization ability.