Application of Improved BP Neural Network in Predicting Teaching Effectiveness of Information Literacy Education in Colleges and Universities

Abstract

This paper takes the evaluation of teaching effect of information literacy education in colleges and universities as the research object, and establishes the evaluation index system of teaching effect of information literacy education, which includes 13 first-level indexes and 30 second-level indexes in the four dimensions of students' information consciousness, information application, information ethics, and information ability. The BP neural network improved by particle swarm optimization algorithm is used to establish a mathematical model for evaluating the teaching effect of information literacy, which uses the BP model trained by PSO to fit the complex relationship between the many indicators affecting the evaluation of the teaching effect of information literacy in colleges and universities and the evaluation results. Through the data research on undergraduates of University D and training analysis using the research data, it is concluded that the training error of PSO-BP algorithm is stabilized at less than 0.02 after 5000 steps of training, and the training time of the model is 10 s. The BP neural network improved by the PSO algorithm is more stable than the pre-improvement connection weight learning process and achieves the ideal error accuracy faster, and its predicted output value is basically in line with the expected value. Its predicted output value is basically consistent with the expected value. The evaluation results also show that the information literacy education in School D needs to be strengthened in terms of information awareness and information application. It shows that the PSO-BP model can be used to effectively predict the effectiveness of information literacy teaching in colleges and universities.