AlphaStar: An Evolutionary Computation Perspective

  • 2019-02-08 15:05:35
  • Kai Arulkumaran, Antoine Cully, Julian Togelius
  • 0

Abstract

In January 2019, DeepMind revealed AlphaStar to the world-the firstartificial intelligence (AI) system to beat a professional player at the gameof StarCraft II-representing a milestone in the progress of AI. AlphaStar drawson many areas of AI research, including deep learning, reinforcement learning,game theory, and evolutionary computation (EC). In this paper we analyzeAlphaStar primarily through the lens of EC, presenting a new look at the systemand relating it to many concepts in the field. We highlight some of its mostinteresting aspects-the use of Lamarckian evolution, competitive co-evolution,and quality diversity. In doing so, we hope to provide a bridge between thewider EC community and one of the most significant AI systems developed inrecent times.

 

Introduction (beta)

None

 

Conclusion (beta)

None