Experimenting with classifiers in biofeedback-based mental effort measurement

Physiological sensors are widely used in order to infer the mental effort of a subject during performing different tasks. Desktop or mobile applications like educational games can gain from such information in order to fine tune the difficulty or the type of a given assignment. Discussions can be found on the advantages and disadvantages of different sensor types (like EEG, ECG, pupillometry, GSR etc.). Machine learning technologies are used to find the baseline of a subject and infer the mental effort levels. In this paper we investigate and compare different types of physiological sensors and classification techniques. Real life experiments with mobile adaptive educational framework (AdaptEd) are presented to support our results.