Alpha Go, Master why go

This is a created article in which the information may have evolved or changed.

To put it simply, Alpha Go, Master is a function y = f (x), x is the current piece position, Y is the x's position, that is, the next move, due to the antagonism of chess, obviously y is the most unfavorable step of X.

Assuming that there are N games, each game has K-step (1<= K <361), when the first step of entering the 1th games, using y=f (x) calculation, y is the alpha Go, master calculated the 2nd step of the Lazi position, because the function f did not learn all the game, So y is not the right way to go, and the 2nd step in the game is biased, the purpose of learning is to make this deviation to the minimum of zero. Alpha Go, Master uses the deep learning algorithm to train this function f continuously, and when the training is done, it will be better for each move.