A DEEP LEARNING-BASED SYSTEM FOR IMPROVING PRIMARY EDUCATION

Abstract

The invention belongs to the field of educational informatics and artificial intelligence and is a deep learning-based system for improving primary education and a method for optimizing personalized learning paths. The system collects data on students' learning and knowledge mastery behavior, models students' knowledge status using a deep learning model, and uses a knowledge graph to construct a dependency graph of knowledge points. The problem of generating learning paths is modeled as a Markov decision process, applying deep reinforcement learning for path finding and global optimization, with integrated consideration of learning benefit, cognitive load, retention of interest, and time constraints, so that dynamic learning paths are generated that meet individual differences.The system updates the dependency graph and the paths during the learning process in real-time, implements adaptive optimization, and displays the recommended paths and their rationale via a visualization interface of the knowledge graph, with support for teacher intervention. (Fig 9).