Re: Commercial video game that learns from its own mistakes?

baylor wrote:

> It's not hard. In fact, i penciled in an article on how to learn from
> mistakes that shows a pretty simple, realistic and CPU-lite way to do

I am very interested in seeing what you came up with.  Ive been looking
at this problem for a while and have been appalled at the difficulty of
doing this for game problem domains over a certain level of complexity.




>
> it. But then my buddy Mike rejected it from the peer reviewed journal
> it was submitted to, so i haven't finished it. Maybe i'll do that and
> post it on my Web site and here. i'm also working on another paper on
> sign tracking and habituation in modality-specific temporal contiguity
> learning (heh heh heh, i used a lot of big nonsense words) that might
> be useful, although it's a lower level mechanism type thing
>
> -baylor