Multimodal emotion recognition algorithm based on edge network emotion element compensation and data fusion

The data feature set of emotion recognition based on complex network has the characteristics of complex redundant information, difficult recognition and lost data, so it will cause great interference to the emotion feature of speech or image recognition. In order to solve the above problems, this paper studies the multi-modal emotion recognition algorithm based on emotion element compensation in the background of streaming media communication in edge network. Firstly, an edge streaming media network is designed to transfer the traditional server-centric transmission tasks to edge nodes. The architecture can transform complex network problems into edge nodes and user side problems. Secondly, the multi-modal parallel training is realized by using the cooperative combination of weights equalization, and the reasoning of nonlinear mapping is mapped to a better emotional data fusion relationship. Then, from the point of view of non-linearity and uncertainty of different types of emotional data samples in the training subset, emotional recognition data compensation evolves into emotional element compensation, which is convenient for qualitative analysis and optimal decision-making. Finally, the simulation results show that the proposed multi-modal emotion recognition algorithm can improve the recognition rate by 3.5%, save the average response time by 5.7% and save the average number of iterations per unit time by 1.35 times.