COSLETS: Recognition of Emotions Based on EEG Signals

In the recent years, one of the leading technology which is earning greater mode of interest in the growing various fields of artificial intelligence is Brain computer interfaces (BCI). Recognizing emotions based on physiological signals specifically, Electroencephalography (EEG) signals with advancement of BCI applications, has turn into a very popular research topic. In this paper for effective representation of features the proposed model adopts COSLETS transformation approach, a combination DCT (Discrete Cosine Transform) and wavelets Transform. The obtained set of features is mapped on to the low dimensional subspace to employ principal components using PCA and finally GRNN (General Regression Neural Network) is presented for effective classification of four different emotional states from publicly available EEG based GAMEEMO dataset. The experimental results are promising and performed well, compared to other state of methods.