A Comparative Study of Signal Recognition Based on Ensemble Learning and Deep Learning
Yue Yang, Meili Zhang, Hongmei Pei · 2023
In this paper, based on the traditional artificial feature extraction method, the ensemble learning method and the convolutional neural network method are used to identify the types of 9 kinds of analog and digital modulation signals, and the performance is evaluated and compared. It is found that the convolutional neural network has higher SNR tolerance than the ensemble learning algorithm, and has better classification performance under complex SNR conditions. The ensemble learning algorithm has better classification effect on Chirp and OFDM signals than the convolutional neural network. This research result also provides a research basis for the combination of traditional machine learning methods and deep learning methods.