Cobra Forums

I saw here card game and genetic algorithm. Probably, you can help me. I want to create the self-training program which plays five-card stud-poker. In this game only one closed card. I hope, it will simplify the problem. I read articles, searched for dynamic libraries with neural networks... But I can't understand, how to adapt them for such problem. It is not obligatory, that the program was clever, but it shouldn't make very silly actions. I think, the program should operate in the assumption that the human always make reasonable decisions. Then it can narrow the list of cards from 13 (2, 3 .. King, Ace) to smaller quantity. I am assured that such problem already dared many people. I do not see sense to "invent a bicycle". So, where can I steal such bicycle?

I have not written this exact program. You could look for such a program written in Java, C# or Python and then port it to Cobra.

Before getting into the AI aspect, you could just get the basics working: the deck, the shuffle, the correct turns, a simple "bot" that acts randomly, accepting human player input, etc.

Once you have that established, you can work on whatever AI algos you want to.

If you have further questions or comments, let me know.