Evaluation of university student education management effect based on data augmentation and transfer learning for remote sensing applications

Abstract Evaluating the effectiveness of education management requires the integration of multi-source data and information. Based on data modeling technology, combined with data enhancement and transfer learning methods, this paper analyzes the differences in the allocation of education management resources in six universities in different semesters, and systematically explores the actual effectiveness of university education management. By combining data enhancement technology, we expanded the training data, simulated various real-life scenarios, and ensured that the model is more robust to various data changes. This study mainly used two models: simulation-verification model and BP (back propagation) neural network model, and analyzed their management efficiency, prediction accuracy, stability and time cycle. This study proposed two models: simulation-verification model (evaluating the effect by simulating the consistency of management conditions and verification results) and BP neural network model (prediction model based on data enhancement and transfer learning). Experiments show that the BP neural network model is superior to the simulation model in management efficiency (ratio of resource input to actual effect) and stability (volatility of model prediction results), with an average management efficiency of 85.9%, prediction accuracy of 93.1%, and stability of 72.3%. The BP neural network model is superior to the simulation verification model in terms of management efficiency, prediction accuracy, and stability, demonstrating the potential of integrating advanced data processing technologies such as data enhancement and transfer learning into the education management system.