Reinforcement learning model creates efficient expert learning paths
EVOL: Simulator-Guided Evolutionary Expert Synthesis for Deployment-Free Learning Path Recommendation
Artificial Intelligence
Summary
Reinforcement learning struggles to recommend sequences of learning steps because it's hard to get feedback until the end and real expert paths aren't in the data. The authors used a virtual simulator that mimics student learning to generate expert-like learning paths by evolving solutions. They then trained a model to follow these expert paths, even without direct feedback during use. Their method named EVOL performed better than eight other techniques on three different tutoring datasets.
What this means in practice
- •For education technology developers: Create personalized learning recommendations without relying on expert-labeled sequences by using simulator-based evolved expert demonstrations.
- •For adaptive tutoring system builders: Improve effectiveness of automated tutoring by training recommendation policies that plan learning paths under sparse feedback conditions.
Authors
Geonwoo Bang, Dongho Kim, Moohong Min
Abstract
Reinforcement learning (RL) for learning path recommendation (LPR) faces two coupled obstacles. First, the policy must commit to a sequence of L concepts without intermediate feedback, producing a combinatorial search space that grows super-exponentially with L and provides reward only at the final step. Second, expert learning paths would be the natural cure for sparse-reward RL, but they do not exist in educational data, because student logs record what learners did, not what they should have done. We address both obstacles by importing a recipe from simulator-based demonstration learning in robotics: the knowledge tracing simulator is used both to synthesize per-learner expert demonstrations through evolutionary search and to train a deployment-free policy that distills these demonstrations into a feed-forward learner. Our framework, EVOL, instantiates this pipeline with an asymmetric actor-critic where the actor commits to deployment-realistic blind planning while the critic exploits the privileged simulator state during training. Across three datasets (ASSIST15, Junyi, and EdNet; 39-189 concepts) and path lengths L = 5, 10, and 20, EVOL surpasses 8 baselines spanning heuristic, sequential, RL, graph-enhanced RL, and LLM-enhanced methods. We further compare three imitation strategies (BC, AWR, and DAPG) and show that final performance is governed by the quality of evolutionary experts rather than by the particular imitation objective.