Emotion recognition from EEG signal using ISO-FLANN with firefly algorithm

Emotion has a greater importance in communication, perception and decision making. The ability to identify the emotional states of people surrounding us is an essential part of natural communication. Human brain activity is used to analyze the emotions in different situations and brain activity is measured by EEG signal. The existing system has concluded a 6 layer biologically tested feed forward neural network which is used to extract the human emotions through EEG. This system delivered to a series of functionalities such as preprocessing, feature extraction, classification. In first phase the EEG preprocessing is done by band pass filter. The connectivity feature is estimated by magnitude square coherence estimation (MSCE) in second phase. In third phase nonnegative sparse principal component analysis (NSPCA) and radial basis function are used for feature selection and classifications respectively. In order to achieve improved accuracy and high true positive rate the proposed system introduced a Firefly algorithm for feature selection and ISO-FLANN for classification. In this proposed system, optimal features are selected by using accuracy as an objective function. Finally the optimally selected features are then classified using an ISO-FLANN to discriminate various states of emotions. From a practical observation it conveys that the proposed system achieves higher accuracy when compared to the existing system.