The success and wider adaptability of smart phones has given a new dimension to\nthe gaming industry. Due to the wide spectrum of video games, the success of a particular\ngame depends on how efficiently it is able to capture the end users� attention. This leads to\nthe need to analyse the cognitive aspects of the end user, that is the game player, during game\nplay. A direct window to see how an end user responds to a stimuli is to look at their brain\nactivity. In this study, electroencephalography (EEG) is used to record human brain activity during\ngame play. A commercially available EEG headset is used for this purpose giving fourteen channels\nof recorded EEG brain activity. The aim is to classify a player as expert or novice using the brain\nactivity as the player indulges in the game play. Three different machine learning classifiers have been\nused to train and test the system. Among the classifiers, naive Bayes has outperformed others with\nan accuracy of 88%, when data from all fourteen EEG channels are used. Furthermore, the activity\nobserved on electrodes is statistically analysed and mapped for brain visualizations. The analysis has\nshown that out of the available fourteen channels, only four channels in the frontal and occipital brain\nregions show significant activity. Features of these four channels are then used, and the performance\nparameters of the four-channel classification are compared to the results of the fourteen-channel\nclassification. It has been observed that support vector machine and the naive Bayes give good\nclassification accuracy and processing time, well suited for real-time applications.
Loading....