부산대학교 도서관

BP neural network is applied to forward multi-layer backpropagation learning. Its main idea is forward propagation and backpropagation of error in the studying process. The BP neural network has a three-tier structure. Aiming at the problem of cultivating English majors' labor consciousness, we propose a research method based on BP neural network. Firstly, 2433 questionnaires were collected. According to the questionnaire survey, two approximate primary indexes and five detailed secondary indexes are set to train and test BP neural network. The accuracy, recall, and AUC values of the trained BP neural network were evaluated. The BP neural network had high accuracy, recall, and AUV values, which were 0.962, 0.702, and 0.6456, respectively. This indicates that the BPNN model was reliable and accurate. We further study and compare BPNN with the Decision tree model and SVM. With calculation, analysis, and comparison, the results show that BPNN has high accuracy in various indicators, and the recall rate is 30.9% higher than that of the Decision tree model, which greatly improves the accuracy of the model and provides great advantages.