Attention guided EEG based Game Genre Classification

Video games are ingrained in people of all ages and backgrounds demographically. The primary objective of a game is to amuse the player, and the designer's fundamental game design concepts have a direct impact on the player's enjoyment. In this paper, a framework is presented in which electroencephalography (EEG) is used for classifying the two-game genres - Action and Puzzle, based on player cognitive attention. The features used for game genres classification are frequency domain characteristics (i.e. band average power) and Sample Entropy derived from EEG signals of gamers. Based on the features it is observed that a player's brain activity significantly varies when interacting with the different game genres. The machine learning models - Support Vector Machine (SVM) and Neural Network (NN) is employed for classification. Conclusively both techniques effectually model the EEG signal recorded during game genres play. However, under an optimal selection of parameters, the Neural Network (NN) model outturn the highest possible accuracy of 98.35% as compared to the Support Vector Machine (SVM) which attained the utmost accuracy of 96.24%. It is also observed that Neural Network (NN) classifies the game genres more effectively in comparison with Support Vector Machine (SVM).