Explore a research paper on Reinforcement Learning in Real-Time Strategy (RTS) games in this 23-minute video. Delve into the insights provided by an ablation study that sheds light on crucial elements of machine learning models like OpenAI Five and DeepMind's Alpha Star. Examine the paper "Gym-μRTS: Toward Affordable Deep Reinforcement Learning Research in Real-Time Strategy Games," which aims to make deep reinforcement learning research more accessible in RTS games. Learn about the environment, results, invalid action masking, and other key takeaways from this study. Gain a deeper understanding of what makes these advanced ML models work and the mysteries that still surround their functionality.
Overview
Syllabus
Intro
Environment
Results
Invalid Action Masking
Minor Takeaways
Taught by
Edan Meyer