AlphaGo: Google's program for playing Go

Datum konání: 18.03.2016
Přednášející: Jan Kotera
Odpovědná osoba: Kotera

Reflecting the recent events in our field which flooded even non-technical media, I will describe the principles of Google's Go playing program AlphaGo, which uses a combination of deep neural networks and "classical" Monte Carlo tree search and is the first computer program to ever beat a top-playing human in the full-sized game of Go.