Real-time detection of emotional changes for inhabited environments

The utilisation of emotional information in computer systems that interact with humans has become more prevalent during the last few years. The various channels through which emotions are expressed provide valuable information about the way humans think and behave and have been successfully employed to assist the inference mechanism of interactive computer applications. In this paper a novel approach to detect changes in the emotional status of a subject is presented. It is argued that the proposed methodology will be able to detect emotional changes in real time utilising physiological measures and a combination of Artificial Neural Networks (ANNs) and statistical mechanisms. Clustering analysis is used to show that the myogram signal was the most suitable attribute to distinguish between two emotional states. Results show that the suggested mechanism is able to accurately distinguish changes from neutral to non-neutral emotional states. Emotional information could be employed to improve user interaction in inhabited environments.