Paradigm Shift in Chess Engines

MLConference, Munich, June 2019

Oliver Zeigermann / @DJCordhose

http://bit.ly/mlconf-chess

Wrap Up

Combining state of the art AI techniques can bring us a long way

• chess playing has been dominated by alpha-beta search for decades
• compute power and hand-crafted evaluation functions are key
• new paradigm
  • Deep Neural Network instead of hand-crafted evaluation function
  • Monte Carlo Tree Search instead of alpha-beta search
  • Self play and learning using Reinforcement Learning
• we see successes and an overall new way of playing

Paradigm Shift in Chess Engines, MLConference, Munich, June 2019

Oliver Zeigermann / @DJCordhose
http://bit.ly/mlconf-chess