M25.6 CONNECT THE MECHANISM
Learn when other decision-makers are learning too
Two booking assistants chase the last hotel room and both end up paying more. Learn to read the game, and see why AlphaStar trained a whole league of rivals.
LESSON OVERVIEW15 min lesson
Lesson overview
Two booking assistants chase the last hotel room and both end up paying more. Learn to read the game, and see why AlphaStar trained a whole league of rivals.
What you’ll explore
- Game-theoretic and multi-agent learning settings model interacting objectives; self-play can generate useful experience while nonstationarity and strategic overfitting complicate evaluation.
GO TO THE SOURCE
Original explanations, connected to the research.
Multi-Agent Actor-Critic for Mixed Cooperative-Competitive EnvironmentsArtificial Intelligence: A Modern Approach — authors’ materialsEquilibrium points in n-person games (Nash, PNAS 1950)Mastering the game of Go with deep neural networks and tree search (Silver et al., Nature 2016)A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play (Silver et al., Science 2018)Dota 2 with Large Scale Deep Reinforcement Learning (Berner et al., 2019)Grandmaster level in StarCraft II using multi-agent reinforcement learning (Vinyals et al., Nature 2019)Suggest a correction
A precise note can make an explanation better.
Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.