Abstract:
Research on spectrum behavior analysis of cultural group communication has gradually changed from the traditional method based on manual feature extraction to the intelligent method based on deep learning. In this paper, a communication behavior analysis model of cultural groups is established based on deep belief network algorithm. In this paper, the constrained Boltzmann machine in each layer is pretrained by layer-by-layer training, and the weight and bias parameters are updated by mapping between the multi-layer constrained Boltzmann machines. Then this paper uses deep belief network to fine-tune the updated parameters. Simulation results show that the proposed method can effectively improve the accuracy and speed of cultural group communication anomaly data acquisition. © 2023 IEEE.