Multiscale Feature Extraction Techniques for Emotion Recognition from Physiological Signal

Emotion detection from physiological data is a growing field with important uses in affective computing, HCI, and medical imaging. To achieve significant progress in human-computer interaction, ability to recognise emotions is essential, particularly when using non-intrusive physiological sensors. This work investigates multiscale feature extraction methodologies used in emotion detection systems, with a focus on methodology and benefits. We examine physiological signals like electrocardiogram (ECG), galvanic skin response (GSR) & electroencephalogram (EEG) and we discuss state-of-the-art multiscale feature-based deep learning and machine learning methods. The aim of the paper is to present thorough assessment of state-of-the-art multi-scale machine learning techniques.