The well-known artificial intelligence

[[email protected]]

Although the rules of the game of Go are simple, the game process is very complex. Behind every decision made by the players there is a possibility of " 2 x 10 170 " . Experienced chess players learn through years of trial and error, and this decision-making mechanism is called reinforcement learning .

If tens of millions of match data between professional belarus whatsapp number data 5 million chess players are input into artificial intelligence, what kind of learning results will it ultimately produce? The answer is AlphaGo , Go software that defeated top human players. Its operating principle is based on deep learning.



But what if we change the approach and let the artificial intelligence understand the rules of Go first, and then let it simulate millions of chess games " on its own" to learn how to play the game? In addition to allowing artificial intelligence to create its own data (millions of chess games), "deep reinforcement learning" can also help it find the best move by analyzing different positions. Just like the human learning model, artificial intelligence will adjust subsequent actions based on its experience of success or failure in order to improve the final result. However, the scope and thinking speed it can cover when making decisions are far beyond human capabilities. and realm.