Issue Details

RECOGNITION OF MULTIMODAL EMOTIONS WITH A HIERARCHICAL NEURAL NETWORK

ABSTRACT

Deep learning together with electro-encephalograms has been extensively used in recent years in the field of multimodal emotional recognition. Because of the complexity of electroencephalograms, some scientific researchers have employed profound education to uncover new parts of emotional detection. In previous experiments, the neural model network was used to automatically extract functionalities and fully recognise the feeling and achieve particular results. However, it is still being studied with a convolutionary neural network to extract hierarchy features. The paper therefore proposes hierarchical neural network fusions to explore data possible data information through the building of diverse structures in the hierarchy of the network, the extraction of multi-scaling features and the use of fusion to combine weights with statistical features manually extracted to produce the final vector characteristics. In this study, the DEAP and MAHNOB-HCI data sets’ valence and excitation dimensions are examined in binary classifications in order to test the model’s effectiveness. The results show that 84.71% and89.00% of the two data sets were precise with the proposed model, which indicates that the model provided in this study was higher than prior models for classification of deep emotional learning in function extraction and fusion.

FULL TEXT