December 2005


Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Peter Drake <[log in to unmask]>
Reply To:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Fri, 30 Dec 2005 10:42:03 -0800
text/plain (37 lines)
On Dec 29, 2005, at 2:58 PM, Liviu Panait wrote:

>> I set the minimum to zero in the belief that fitnesses have to be
>> nonnegative.  Is this true?
> SimpleFitness may have arbitrary values as of the previous version (I
> believe, or if not the current version does not complain about that
> anymore).

Yes, I guess there would be no need to "normalize" the values if we
use tournament selection.

> I am not sure I follow the reasoning.  Even if an individual won
> against early competitors in initial generations, that says little
> about how good he is against better opponents found at later
> generations.  I would treat this on a per-generation basis and ignore
> information about wins or loses from previous generations.

I had thoughts about accumulating fitness information so I didn't
have to play a lot of games each round, but in retrospect I see that
this is already taken into account by the GA.

> Again, I encourage you to look into the single-elimination
> tournament, which we found to work nicely.  There is also a version
> of double-elimination-tournament somewhere, though we have not tested
> it that much.

I read your GECCO paper, and I'm convinced by your argument.  Back to
the bit mines with me!


Peter Drake
Assistant Professor of Computer Science
Lewis & Clark College