AlphaGo Zero

Terms from Artificial Intelligence: humans at the heart of algorithms

AlphaGo Zero is the successor of AlphaGo, which defeated a Go grandmaster for the first time in 2016. Whilst AlphaGo was initialy trained on human games, AlphaGo Zero started from scratch with only a description of the rules of Go and thereafter learning entirely using adversarial training.

Used on page 238