Human emotional understanding has become one of the most rapidly emerging fields developed over the years since the emotional state is created and influenced by several neither easily measured and nor predicted factors. The proposed novel approach is based on the mixture of the processed acceleration data of several designed activities and processed heart rate data under several emotional states, recorded by the application of`smart band devices. Since the processed acceleration and heart rate data are running on different dimensionalities and recorded by different types of sensors, some machine learning-based methods applied to fusing these two kinds of data. The results of the data fusion methods are defined as the new fused datasets for Multi-Layer Perceptron and Support Vector Machines classifiers based on a selected criterion. At last, the accuracy results of the selected classifiers running under the defined machine-learning-based fusion dataset compared to each other to find out the effectiveness of the proposed fusion method.
