Affective computing has become an increasingly popular area of study that focuses on developing a robust system that automatically recognizes human emotions. Any changes in human emotion directly impact brain stimulation and physiological parameters. Emotions are critical parameters that directly impact the behaviour of an individual. Collecting the neurophysiological data from humans is utilized here. The ultimate goal of the presented work is to develop a robust methodology to predict human real affect. The present system considers the Amigos data set in which physiological signals such as ECG, EEG, and Galvanic Skin Response (GSR) are considered. The proposed framework is modelled using Gaussian mixture models to produce an expectation-maximization technique (GEM). Along with the measurement of statistical parameters such as mean, standard deviation, and Sigma, these are helpful for the system to identify the real class. The comprehensive analysis of various participant data collected through the AMIGOS dataset was split into training data and testing data. The raw data preprocessing using the Synthetic Minority Oversampling Technique (SMOTE) model in which the data has cleaned up random values or removed junk values or removed. Since data is provided to the GEM algorithm in which the expectation-maximization parameters such as mean median standard deviation and sigmoid are calculated. Based on the evaluated model, part of the test signals are given from the amigo dataset. The system can compare an interpreter to validate the parameters-based emotion labelling. A novel EmoNet_ANN (ENA) system is further improved by including deep learning models to find the detailed covariate values helpful to identify the personality traits of the participants. The proposed optimized Novel EmoNet_ANN (ENA) perceives the accuracy benefit of 92.5% with less computational time. A novel EmoNet_ANN(ENA) Proposed system is comparatively studied and further improved by evaluating deep learning models to find the detailed covariate values present in the data set. The statistical measures on Sigma, Variance from the GEM model concerning the correlation metric from ANN are further used to validate the emotion detection. Accuracy, Precision, Recall and F1score form the analysis module which is used to determine the overall performance statistics. Emotion analysis and affect impacted parameters are modelled using Novel EmoNet_ANN (ENA). The AMIGOS dataset analyses physiological signals to determine the actual emotions. With reduced computation time and iteration run till Zero error on complete analysis.

No abstract