Originally published in IEEE Spectrum on December 23, 2020
By Philip E. Ross
DeepMind after having introduced a reinforcement learning system named AlphaGo in 2016, AlphaZero in 2018 has recently introduced MuZero that does not require to be taught the game rules. MuZero attempts different alternatives one after the other and forms the strategies on the fly until it collects the rewards more quickly. MuZero due to its amazing capabilities has outperformed the previously developed systems like AlphaGo and AlphaZero that required training on games.
To read more Tech News, click here.