From Tic Tac Toe to AlphaGo - Playing Games with AI and Machine Learning

Explore the groundbreaking journey of AI in game-playing, from simple Tic Tac Toe to the complex world of Go, in this 50-minute Devoxx conference talk by Roy van Rijn. Delve into the algorithms and techniques that enabled Google's AlphaGo to achieve its extraordinary breakthrough, beating top human players in a game with 1.74×10^172 unique positions. Learn about chess algorithms, the horizon problem, pruning techniques, and the transition to tackling Go's complexity. Discover the power of Monte Carlo Tree Search, neural networks, and reinforced learning in AI game-playing. Gain insights into the development of AlphaGo, its matches against human champions, and the subsequent emergence of AlphaGo Master. Conclude with a discussion on the potential applications of these AI advancements in healthcare and other fields, followed by a Q&A session.

Intro
What do you need
Chess
Problem with Chess
Horizon Problem
Pruning
Pruning is dangerous
Better pruning
Minimax pruning
Chess bot
Chess branching factor
Go
Problems with Go
Monte Carlo Tree Search
How long will it take
Neural networks
Neural network playground
Tensorflow
convolutional neural networks
Reinforced network Go
Value network Go
Combining all the pieces
The Challenger
The Challenger Game 1
The Challenger Game 2
The Challenger Game 4
Conclusion
Nobody thought AlphaGo
AlphaGo went silent
New player MasterP
Healthcare
Questions