Information générale
Catégorie:
Fonctions Windows API
I've got a rough schema for pattern recognition vai a neural net type arrangement that can be stored in .dbfs. It's too crude to show at the moment (it works though!). But it's one of my retirement projects!
In game AI the key thing is to see "patterns" and make decisions based the "score" (or scores) of that pattern, perhaps relative to known patterns. There's no way, in anything but the most simple of games, that you can describe in procedural code how to make a "clever" decision. But of course there's much much more...
The other thing that I've learn't is that a "blank" neural net is not much good. It simply takes too long to train. You really have to purpose build the "net" with much of the heuristic (rules of thumb) level of intelligence pre-built. That's hard :-)
anyway c u round!
Dave
>Yeah, I see where you're going! I wrote a chess/othello type game not too long ago in a Basic/C hybrid language and have been wanting to rewrite it in VFP also. The problem I had was that as a two human player game it was easy to write but when I started writing the AI for it, the code for the AI became longer than all my other code for the whole game. Hence, I never finished the one player/computer option in the game. I was thinking that it would be cool to write a game that the computer actually "learned" how to play better. So the computer would really be an equal partner in that as you got better at the game so did the computer. I know this isn't a "new" approach, but for me it would be fun to write something like that. The cool thing about VFP like you mentioned is that you could use tables to STORE the AI as it grew. If it really "learned" the players techniques then even trading saved AI's with other players would allow people to play other people's AI's that had learned their
>particular techniques.
Précédent
Suivant
Répondre
Voir le fil de ce thread
Voir le fil de ce thread à partir de ce message seulement
Voir tous les messages de ce thread
Voir tous les messages de ce thread à partir de ce message seulement