ECJ-INTEREST-L Archives

January 2010

ECJ-INTEREST-L@LISTSERV.GMU.EDU

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

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

Print Reply
Sender:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Date:
Thu, 28 Jan 2010 09:04:04 -0500
MIME-version:
1.0 (Apple Message framework v936)
Reply-To:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Content-type:
multipart/mixed; boundary=Apple-Mail-11-764919385
Subject:
From:
Sean Luke <[log in to unmask]>
In-Reply-To:
Comments:
To: ECJ Evolutionary Computation Toolkit <[log in to unmask]> cc: Vlad Palnik <[log in to unmask]>
Parts/Attachments:
text/plain (457 bytes) , SimpleBreeder.java (11 kB) , text/plain (11 kB)
Evaluation is performed on any individual whose "evaluated" flag is  
false.  So you could go through all of the population after breeding  
and set 'em all to false.  That's ugly though.  The right thing should  
have been for that feature to be in SimpleBreeder all along.

I've committed a new version of SimpleBreeder which, in my delerious  
state, appears like it SHOULD work, but is utterly untested.  I'm  
attaching it here.  It's your job to test it.  :-)




It's got a new parameter which you can set to specify whether a given subpopulation's elites should be reevaluated or not. When it sees that it flags them as not evaluated when they're cloned into the new subpopulation. Let me know if it appears to be working. Sean On Jan 28, 2010, at 12:09 AM, Vlad Palnik wrote: > Hi all, > > I am looking for the code that bypasses the reevaluation of elites. > The individuals in my problem set are evaluated using a different > data set at each generation. I would like to force the reevaluation > of elites and assign new fitness values relevant to the specific > data set. I've been trying to find the code responsible, but no luck > so far, SimpleBreeder does not reference evaluated, and I have tried > to bypass fitness.clone() in Individual during elite cloning with no > results. Does anyone have any ideas??? > > Vlad

ATOM RSS1 RSS2