Training an Intelligent Tutoring System Using Reinforcement Learning

Abstract: In this work we have applied reinforcement learning in building an intelligent tutoring system. An intelligent tutoring system is a computer system that provides personalized learning material to the learner, based on his needs and level of knowledge. Such a system may consist of the following components: the knowledge base, the student’s model, the pedagogical module and the user interface. The role of the pedagogical module is to define what is the best learning material to give to the students in order to help them reach their goal towards learning the material of the course. This is a continuation of our previous work that models an intelligent tutoring system as a reinforcement learning problem for teaching different lessons related to Python programming language [1]. In this work, we focus on building the pedagogical module through applying reinforcement learning and the DQN algorithm. To model a problem as a reinforcement learning problem, we should take special care in defining the following components: the state space, the actions and the rewards. Here, we propose a way to organize the state space, the actions and the rewards, in order to train the pedagogical module using reinforcement learning. After defining those elements, we train this module using different parameters and conditions. The training is done in a simulated environment, by simulating the behavior of the student in order to help the training process.
Keywords: intelligent tutoring system, reinforcement learning, DQN