Classification of Human Emotion from Deap EEG Signal Using Hybrid Improved Neural Networks with Cuckoo Search

Abstract

Department of Computer Science and Engineering,
Anna University Regional Centre, Coimbatore, India
m.sribtechit@gmail.com
J. Preethi
Department of Computer Science and Engineering
Anna University Regional Centre, Coimbatore, India
preethi17j@yahoo.com
Emotions are very important in human decision handling, interaction and cognitive process.  In this paper describes that recognize the human emotions from DEAP EEG dataset with different  kind of methods. Audio — video based stimuli is used to extract the emotions. EEG signal is divided  into different bands using discrete wavelet transformation with db8 wavelet function for further  process. Statistical and energy based features are extracted from the bands, based on the features  emotions are classified with feed forward neural network with weight optimized algorithm like  PSO. Before that the particular band has to be selected based on the training performance of neural  networks and then the emotions are classified. In this experimental result describes that the gamma  and alpha bands are provides the accurate classification result with average classification rate of  90.3% of using NNRBF, 90.325% of using PNN, 96.3% of using PSO trained NN, 98.1 of using  Cuckoo trained NN. At last the emotions are classified into two different groups like valence and  arousal. Based on that identifies the person normal and abnormal behavioral using classified  emotion.