AMIGOS: a robust emotion detection framework through Gaussian ResiNet

Abstract

Affective computing is the study of the deep extraction of emotional impacts that triggers humans for various reasons. Emotions directly reflect on human behaviour. The proposed analysis is inclined towards deep emotion extraction through a novel concept with less computation time. Designing a robust analysis model is focused on here. AMIGOS dataset on affect, and personality modelling is considered here. A novel gaussian ResiNet (GRN) algorithm is evaluated here. Any changes in the emotions of humans are the brainy response given to the actions faced. The features of the given physiological factors are considered for analysis, further with GMM-ResiNet (GRN) a low computational structure is used for classification. The novel gaussian ResiNet (GRN) is created from the given dataset for similar feature validations. The system predicts the correlated relative data from the training set and testing set and achieved the performance metrics using error rate (ER), algorithm computation time (ACT), full computation time (FCT), and accuracy (AUC). novel gaussian ResiNet (GRN) is created and tested with processed data of the AMIGOS dataset. The model created is validated with state-of-art approaches and achieved an accuracy of 92.6%.