Discrete Emotion Classification Based on Skin Conductance: A Comparative Study of Machine Learning Models Across Contexts

Abstract

With the rapid development of the field of affective computing, the accurate identification and classification of human emotions have become increasingly important. This study aims to classify neutral emotion (neu) and other emotions using peripheral physiological signals such as Skin Conductance Response (SCR), Skin Conductance Level (SCL), and Instantaneous Skin Conductance Response (iSCR) through Support Vector Machine (SVM), Multilayer Perceptron (MLP), and Convolutional Neural Network (CNN) models. The study compares the classification effects under different emotion induction contexts (imaginary and video contexts) and evaluates the generalization ability across these contexts. A total of 43 subjects were included in the study, and the accuracy of emotion recognition and the generalization of models were explored by analyzing the physiological signal characteristics in different emotional states. By comparing the classification effect in different emotion-inducing situations, the research helps to find more accurate emotion recognition models, enhance the generalization ability of models, and promote the development and application of emotion computing technology.