The bond between chess and computer is as old as the computer. Innovators like Babbage, Turing and von Neumann worked to create hardware and Algorithms to study and play chess. It is also the most studied game in the history of Artificial Intelligence.
> Google DeepMind Awesome Creation - AplhaZero Algorithm:
Google's DeepMind has a brilliant track record in terms of beating humans in board games. It does so with a long-term goal to create something which can be useful to a wide range of situations. In a new paper, showing his latest work, DeepMind defined how the Artificial Intelligence program learned to play chess and a couple of other games.
alphazero ai, alphazero, alphago, deepmind, google, ai, artificial intelligence, no humans needed
 |
| alphazero ai, alphazero, alphago, deepmind, google, ai, artificial intelligence, no humans needed |
> Only 4 Hours Training and AlphaZero Gives Marvellous Performance:
With only 4 hours of training, AlphaZero AI from DeepMind has settled an exceptional performance in chess. After a "Self-Learning Reinforcement Learning" of 300k steps, he passed the “stockfish”, the best chess program in the world. It is worth noticing that Artificial Intelligence was only set with chess rules and did not fuel game tactics.
> AlphaZero Algorithm vs AlphaGo Zero Algorithm:
The AlphaZero Algorithm is an overall version of the AlphaGo Zero algorithm used to play Go.
About the results of 100 Matches against the stockfish, AlphaZero has “won 25 matches” as a white. With black, he won 3 games. The rest of the games were drawn, and the stockfish could not record a single win.
> AplhaZero Training - See Below How This System Trained?
For training, AlphaZero used a single machine with “4 TPUs”. Stockfish played at their peak skill level using “1 GB Hash and 64 threads”.
0 Comments