I've actually thought about taking it a bit further and generating code with tests as the fitness data, but with so many possible nodes... it could get messy. Heckle may be able to handle it at some point though. I've got the mutation code written! It'd just need to handle cross-overs.
Kev On 12/8/06, Patrick Crowley <[EMAIL PROTECTED]> wrote:
I saw a pretty sweet demo at BarCamp LA that used these techniques. Imagine drawing a simple shape. Then imagine using evolutionary code generation to cycle through thousands of generations of code to find the exact algorithm that reproduces your shape using LOGO commands. -- Patrick On Dec 8, 2006, at 9:06 am, Tom Werner wrote: > That would be fantastic! I would love to see a talk on DRP, I've > been meaning to check it out. So, can we put you down for a talk > next month (Jan 5th)? > > Tom _______________________________________________ Sdruby mailing list [email protected] http://lists.sdruby.com/mailman/listinfo/sdruby
-- Kevin Clark http://glu.ttono.us _______________________________________________ Sdruby mailing list [email protected] http://lists.sdruby.com/mailman/listinfo/sdruby
