EEG-based attention recognition

Attention recognition (AR) is an essential component in many applications, however the focus of current research into AR is on `face detection´, `eye center localization´ and `eye center tracking techniques´. This paper describes a research project conducted to investigate the use of electroencephalography (EEG) signals to extend the current approaches and enrich AR. EEG processing and classification algorithms are applied to EEG data to identify a group of features that can be used to effectively implement AR. The experimental results reported in this paper are encouraging with correct classification rates achieved being: 51.9% where attention is divided into 5 classes and 63.9% where attention id divided into 3 classes. The distribution of the `training tuples´ and `testing tuples´ is discussed along with their impact on the reported results. The paper concludes with an overview of outstanding issues and consideration of projected future research.