Deep Reinforcement Learning Based Game Decision Algorithm for Digital Media Education

On the traditional digital media education platform, problems raised by users are usually solved by searching the optimal solution in the existing database, which lacks the ability of autonomous learning and long-term planning. This paper introduces a Deep Reinforcement Learning algorithm based on Deep Deterministic Policy Gradient to solve these issues. We applied a trained neural network with this algorithm to play "Klotski" - a sliding block puzzle where the aim is to move a specific block to some predefined location. In this article, we explain why the game is selected and how our method works. We also discuss possible ways forward within the field in the future.