AlphaGo Zero
Added to the Museum of Data by Yingqi Zhao on Monday, October 19, 2020. Museum of Data Collection ID: 585.
Public description: AlphaGo Zero is an artificial intelligence software program developed by DeepMind in 2017. AlphaGo Zero is the first program to achieve superhuman performance in the game Go. Unlike other series of AlphaGo programs, AlphaGo Zero's algorithm based entirely on reinforcement self-learning. Winning a game like Go used to be believed that require a complex of analyzing to predict, to have instinct, and memorizing the great amount of experience developed by the predecessors. Therefore, the coming out of AlphaGo Zero is oftentimes perceived as a major breakthrough in the history of artificial intelligence technology.
Materials used: a hardware, and a neural system software
Credit: The research, ethics, science, and operation team of DeepMind. Private company owns commodities, also provided for AI research use to the institutions.
Size: a 4 TPUs single machine hardware
Creation date: 2017/10/19 00:00:00
External link: https://deepmind.com/blog/article/alphago-zero-starting-scratch
Tags: 8, 14