Episode

New reinforcement learning methods aim to improve AI in gaming and beyond

We discuss reinforcement learning – an AI approach that is especially promising for training naturally behaving game characters with nuanced reactions to human players. Viewers learn about key concepts of reinforcement learning in the context of Project Malmo – which enables RL research in the game Minecraft, and Project Paidia – which aims specifically at creating game agents that learn to genuinely collaborate with human players.

Jump To: 

  • [00:50] What is reinforcement learning 
  • [02:26] Project Malmo – Minecraft as a platform for AI research
  • [05:00] Project Paidia: towards RL agents that learn to collaborate with human players
  • [09:00] Project Paidia demo
  • [11:00] Call to action -AI Innovation page (Project Paidia) + Azure ML notebooks 

Learn More: 

Don't miss new episodes, subscribe to the AI Show