At the moment I am writing chess in C ++ Qt. Graphic shell implemented. There was a question about how best to implement artificial intelligence. I imagine it as a bust of all possible moves. I found the information that it is necessary to calculate the options of moves through the Neural Networks. Please give a reference and if possible the material where a similar problem is stated.

  • I would start dancing from a Wikipedia article about Deep Blue with a focus not on historical facts, but on the theory of chess, also touched upon in this article. Generally speaking, there are many materials on this subject, to put it mildly. - Danatela
  • @Danatela I read a series of articles on the design of artificial intelligence in general terms. where is pretty trivial problem. About the deep blue thanks, read. - ParanoidPanda

2 answers 2

Computer toys-chess were certainly already in the 80s ... Therefore, I think it’s unlikely that neural networks are used there, the usual brute force ...

Suppose we play for white, computer for black. The algorithm will be like this ...

  1. Calculate possible moves of black (only a maximum of 16 pieces)
  2. Calculate the possible moves of white
  3. We do so several iterations

The number of iterations is determined by the "computer brain".

We do several iterations, we calculate how many pieces White has on points. The fewer the points, the better ...

GO!

  • Thanks, you really helped, sent my thoughts in the right direction. - ParanoidPanda
  • If anything, I can participate in the project - it became interesting! Write to contacts ... - sitev_ru
  • Sorry, but this is a practical work for the department, so I'd rather grind it myself, but thanks for responding! - ParanoidPanda

see gnu chess (open source). In general, the interface is one hundred billion in chess