EEG SIGNAL-BASED EMOTION RECOGNITION USING A DEEP LEARNING APPROACH WITH GRU-BGWO

A voluntary or involuntary response to outside stimuli is referred to as emotion. Individuals use body language, words, noises, and facial expressions to convey their feelings. However, genuine feelings cannot always be communicated; people might occasionally alter the emotions displayed in such activities. As a result, it is crucial to comprehend and analyse emotions. Because of the more trustworthy data gathered, investigations on emotion recognition based on EEG signals seem to be in the spotlight. This work developed a deep learning algorithm with an optimizer and analysed emotion using EEG data. There are five phases of the investigation. The suggested framework first preprocesses the data using low-pass filtering to eliminate noise before extracting the delta rhythm. Following the signal processing phase, namely the feature extraction phase, two popular technologies were employed: Higuchi's fractal dimension (HFD) and entropy. Deep feature selection using ResNet50 yields the best features, which are subsequently categorized using the Gated Recurrent Unit with Binary Gray Wolf Optimizer (GRU-BGWO) method. To evaluate the efficacy of our suggested approach, experiments were carried out on the publicly accessible DEAP database using a variety of evaluation criteria, such as accuracy, specificity, and sensitivity. The results showed that the proposed method was effective, as the GRU-BGWO-based method achieved 91.45 percent accuracy.