Optimal Hierarchical Learning Path Design With Reinforcement Learning.

Area of Science:

• Artificial Intelligence in Education
• Cognitive Science
• Educational Technology

Background:

• E-learning systems offer adaptive and efficient learning experiences compared to traditional classrooms.
• A critical component of e-learning is the learning policy, which dictates personalized learning paths.
• Current learning policies often lack optimal strategies for complex skill hierarchies.

Purpose of the Study:

• To develop a novel framework for optimizing learning policies in e-learning systems.
• To address the challenge of determining the best learning path for learners with hierarchical skills.
• To enhance the adaptiveness and efficiency of e-learning through intelligent path selection.

Main Methods:

• Developed a hierarchical skill model to represent learners' knowledge structure.
• Integrated a cognitive diagnosis model to assess various mastery levels of hierarchical skills.
• Applied a model-free reinforcement learning method to find the optimal learning policy without assuming learner transition processes.

Main Results:

• Successfully modeled learners' hierarchical skills and their mastery levels.
• Identified an optimal learning policy that considers the hierarchical nature of skills.
• Demonstrated the framework's effectiveness through simulation studies, showing improved learning path optimization.

Conclusions:

• The proposed framework effectively optimizes learning policies in e-learning environments.
• Hierarchical skill modeling combined with reinforcement learning provides a robust approach to personalized education.
• This research contributes to more adaptive and efficient e-learning experiences.