There is a new master of the Chinese game of Go. Google’s artificial intelligence dubbed as “AlphaGo Zero” defeated the previous leader of this game Google’s AI AlphaGo by 100 games to 0.
AlphaGo previously defeated 18-time world champion Lee Sedol, other world’s best players and world number one player Ke Jie. London-based artificial intelligence company DeepMind – acquired by Google – created both AlphaGo Zero and AlphaGo.
In the paper entitled “Mastering the game of Go without human knowledge” published in the Nature journal, DeepMind researchers demonstrated how AlphaGo Zero defeated AlphaGo.
The main difference between these two AI systems is how they learn to play the game of Go. AlphaGo learned how to play the game by learning from thousands of human amateur and professional players. AlphaGo Zero, on the other hand, did not learn from human players. It, instead, learns the game of Go by playing games against itself.
“The system starts off with a neural network that knows nothing about the game of Go,” DeepMind researchers said in a blog post. “It then plays games against itself, by combining this neural network with a powerful search algorithm. As it plays, the neural network is tuned and updated to predict moves, as well as the eventual winner of the games. This updated neural network is then recombined with the search algorithm to create a new, stronger version of AlphaGo Zero, and the process begins again.”
According to the researchers at DeepMind, AlphaGo Zero’s techniques can be applied to other structured problems, including protein folding, reducing energy consumption or searching for ground-breaking new materials.