Which of the following methods is used to learn alphago?

AlphaGo learning mainly adopts the following methods:

1. Learn a lot about human chess players: This is AlphaGo's original learning method. By learning a lot about human chess players, we can understand and simulate human chess strategies and skills.

2. Self-deep learning: After completing the initial learning, AlphaGo began to enter a complete self-deep learning stage. At this stage, it completely abandons the thinking mode of human chess players, but learns Go independently in its own way through "left-right fighting".

3. Combining the advantages of supervised learning and reinforcement learning: In the self-deep learning stage, AlphaGo combines the advantages of supervised learning and reinforcement learning. Specifically, it forms a strategy network through training, takes the situation on the chessboard as input information, and forms a probability distribution for all feasible positions.

4. Training the value network to predict the self-game: AlphaGo also trained a value network to predict the results of all feasible positions by learning the self-game prediction. The standard is-1 (the absolute victory of the opponent) to1(the absolute victory of AlphaGo).

Through the above learning methods, AlphaGo continuously optimizes and improves its Go technology, and finally reaches a very high level.