The Use of Interactive Mobile Technologies in Higher Education to Recognize Behavior and Develop Self-organizing Instructional Strategies for Deep Learning

Abstract

The widespread adoption of mobile intelligent terminals has accelerated the development of ubiquitous learning paradigms in higher education. However, the cultivation of deep learning in mobile learning environments continues to be constrained by device resource limitations, the difficulty of quantifying higher-order learning behaviors, the rigidity of instructional intervention strategies, and insufficient privacy protection mechanisms. To address these challenges, a closed-loop adaptive framework based on device–edge–cloud collaboration was constructed. Multimodal mobile sensing technologies were employed to collect comprehensive learning interaction data across heterogeneous learning scenarios. Through the integration of model pruning and knowledge distillation techniques, low-power feature extraction was achieved at the edge layer, thereby satisfying both terminal computational constraints and real-time processing requirements. Furthermore, spatiotemporal association rules were integrated with a hypergraph convolutional network to facilitate the accurate mining of higherorder cognitive evolution patterns underlying complex learning behaviors. Reinforcement learning and federated meta-learning mechanisms were jointly incorporated to eliminate dependence on predefined rules, enabling the autonomous evolution of personalized instructional strategies and adaptive generalization across diverse learning contexts. In addition, a lightweight interpretable analytics module was embedded to enhance learners’ acceptance of instructional interventions. Longitudinal comparative experiments were conducted within university courses. A reliable technological paradigm and practical reference are therefore provided for the digital and intelligent transformation of higher education through interactive mobile technologies and for the precise facilitation of deep learning development.