Computing nonlinear features of skin conductance to build the affective detection model

Skin conductance (SC) is one of most important physiological signal, which has been proven to contain reliable affective information. In the paper three kinds of affective status are induced by movie clips in laboratory environments, which are happiness, sadness and fear, and the corresponding affective SC signal is collected by Biopac MP150. After preprocessing the original SC signal, several nonlinear features are computed, which include largest Lyapunov exponent, correlation dimension, approximate entropy, and Hurst exponent. Finally based on these features the classifier SVM is used to setup model of affect detection. The experimental results show it is feasible to build affective model based on SC´s nonlinear features.