Training AI to learn and explain complex game strategies like humans
Learning Explainable Representations of Complex Game-playing Strategies
Artificial IntelligenceMachine Learning
Summary
Playing complex games well means understanding and explaining strategies. The paper shows how to teach AI players to learn game strategies in a way similar to how humans think, by turning game moves into simple step-by-step actions or programs. These learned strategies help AI make good decisions in games like chess and grid puzzles. The researchers demonstrate that AI can learn these programs by just watching game plays.
What this means in practice
- •For game ai developers: Create AI players for complex games who explain their gameplay decisions as understandable sequences.
- •For educational game designers: Build game teaching tools that show human-like strategies learned from gameplay to help players improve.
Authors
Abhijeet Krishnan, Colin M. Potts, Arnav Jhala, Harshad Khadilkar, Shirish Karande, Chris Martens
Abstract
As part of learning to play complex games, human players develop develop abstractions for concepts and strategies of gameplay consistent with game rules to improve their performance. These concepts are applied to explain other players' actions, and to inform their own actions in-game. Understanding other players' strategies is a crucial part of such improvement, but requires time and effort. In this paper, we propose a strategy similar to human cognition for training RL agents to synthesize learned strategies and policies as executable procedures based on sequences of gameplay actions. We present methods to automatically learn such programs to play chess and to solve tasks in a grid-based environment. We show that the learned strategies produce effective actions, and can be learned from gameplay data.