Re: GAs for modifying algorithms

>Have you considered "Inductive Logic Programming"?. You put in a training
>set of positive and negative instances in a relation and it gives you a
>prolog program that covers the data. It works for things like inducing
>ancestor using a given parent relation <snip>:

I am really impressing with functional languages, and I think they will be a hot topic 
in the future (we'll see). My only exposure with them was trying to writing a "text 
adventure" using Prolog as the engine to create the rule set. There's a quick blub if 
you're interested at:
http://bang.dhs.org/if/raif/1999/msg03700.html

I've never thought of combining the idea with GAs. I'll have to think about that some 
more. In the meantime, I'll pick up Tom Mitchell's "Machine Learning" book. Thanks for 
the tip.

Patrick Kellogg
[EMAIL PROTECTED]
http://www.patrickkellogg.com