Simulating Student Interactions with Two-stage Imitation Learning for Intelligent Educational Systems

• DOI: 10.1145/3583780.3615060
• Publication Date: 2023-10-21T07:45:26Z
• Authors: Guanhao Zhao; Zhenya Huang; Yan Zhuang; Jiayu Liu; Qi Liu; Zhiding Liu; Jinze Wu; Enhong Chen
• Keywords: Discriminator; Benchmark (surveying)

The fundamental task of intelligent educational systems is to offer adaptive learning services students, such as exercise recommendations and computerized testing. However, optimizing required models in these would always encounter the collection difficulty high-quality interaction data practice. Therefore, establishing a student simulator great value since it can generate valid interactions help optimize models. Existing advances have achieved success but generally suffer from exposure bias overlook long-term intentions. To tackle problems, we propose novel Direct-Adversarial Imitation Student Simulator (DAISim) by formulating Markov Decision Process (MDP), which unifies workflow training generating alleviate single-step optimization problems. construct intentions underlying complex interactions, first direct imitation strategy mimic with simple reward function. Then, an adversarial learn rational distribution given parameterized discriminator. Furthermore, discriminator pairwise manner, theoretical analysis shows that improve generation quality. We conduct extensive experiments on real-world datasets, where results demonstrate our DAISim simulate whose close real promote several downstream services.