ER-ADENN:Design and Implementation of EEG-based Emotion Recognition using Adaptive Dropout Enabled Neural Network

Emotion recognition is a computer science discipline and field of study that deals with the recognition of human emotions from facial expressions, text, physiological signals, and other forms of behavior. Because human emotions are complex and exhibit a wide range of expressions, emotion recognition is a difficult undertaking. The user experience of human-computer interaction can be enhanced by the application of emotion recognition. Emotion recognition, for instance, can be used to modify a computer system's interface to correspond with the user's emotional state. Emotion recognition is a promising field of research with a wide range of potential applications. As the field continues to develop, emotion recognition is likely to become more accurate and more widely used Emotion recognition based on EEG using adaptive dropout enabled neural network where the data collected from the SEED and DEAP dataset gets preprocessed using the adaptive filtering that utilized or dropped out the data for the enhancement. The time domain feature, Hjorth parameter feature, and frequency domain feature are generated by segmenting the data according to the frequencies of alpha, beta, gamma, and delta. The Adaptive dropout enabled network classifier optimized using the Harmony climbing algorithm developed on combining the Harmony search optimization and Hill climbing optimization performs the classification and boost the convergence and recognized the emotions effectively. The proposed model Harmony climbing algorithm with adaptive dropout enable neural network classifier apply on Deap dataset that enhance the performance and achieve the 94.96% accuracy, 95.05% sensitivity and 94.79% specificity on training percentage 90. On similar ways again the proposed model applies on seed dataset it will enhance the performance up to the level of 93.98% accuracy, 93.48% sensitivity and 94.66% specificity.