Improving Subject-independent Human Emotion Recognition Using Electrodermal Activity Sensors for Active and Assisted Living

Abstract

In Active and Assisted Living environments (AAL), one of the major tasks is to make sure that old people or disabled persons do feel well in their environment. Unfortunately, it is still a difficult task to design a learning system or build a machine learning model which can be trained on a group of subjects using physiological sensors and performs well when testing it on other subjects. This paper proposes a dynamic calibration algorithm which presents promising results for subject-independent human emotion recognition. The goal of the calibration module is to calibrate itself with respect to the features of a new subject by finding the most similar subject in the training data. In order to check the overall performance, this approach is tested using the well-known MAHNOB dataset. The results show a promising improvement based on different evaluation metrics, e.g., sensitivity and specificity.