Classification of Task Difficulty in Free Hand Movements using Brain Computer Interfac

In the current study, we investigate fusion and temporal alignment of modalities coming from continuous hand movement and brain activity in Human computer interaction. When dealing with continuous brain data, right segmentation technique, relating the brain and data from hand movement should be applied. A Matlab GUI was developed where shapes with three levels of difficulty were displayed and subjects had to sketch the shapes. Brain activity of the subjects were recorded using electroencephalogram(EEG) sensors. Two segmentation techniques were used and the brain data was aligned Deltat seconds prior to the data from hand movements. Features from brain asymmetries were used for K Nearest neighbor classification. The experimental and classification results suggest that segmentation using one-third power law with a temporal alignment of 1.4 seconds gives a maximum accuracy of around 97% and is much higher than the segmentation using Speed seg.