[2J1-E-5] Human interface, education aid: education

[2J1-E-5-01] Teaching Reinforcement Learning and Computer Games with 2048-Like Games

〇Hung Guei¹, Ting-Han Wei¹, I-Chen Wu¹ (1. National Chiao Tung University)

キーワード：2048-like games, reinforcement learning, pedagogy

2048-like games are a family of single-player stochastic puzzle games, which consist of sliding numbered-tiles that combine to form tiles with larger numbers. Notable examples of games in this family include Threes!, 2048, and 2584. 2048-like games are highly suitable for educational purposes due to their simplicity and popularity. Numerous machine learning methods have been proposed for 2048, which provide a good opportunity for students to gain first-hand experience in applying these techniques. This paper summarizes the experience of using different 2048-like games, namely Threes! and 2584, as pedagogical tools for teaching reinforcement learning and computer game algorithms. With two classes of graduate level students, the average win rates for 2584 and Threes! reached 96.1% and 93.5%, respectively. The course designs were also well received by students, with 4.21/5 and 4.35/5 points from student feedbacks.