From: pdenboef@hotmail.com   
      
   "the 'dude'"  wrote in message   
   news:407172fe$1@news.unimelb.edu.au...   
   > I am using a RCGA with a HUGE search space. I need to extract a set of   
   rules   
   > (say 15-30 rules) from very complex data containing 24 attributes. The   
   > complex data is gathered from a well-trained ANN, which I wish to use in   
   > order to study knowledge representation/ rule extraction.   
   >   
   > Each antecedent has roughly 10 possibilities, which was reduced using an   
   > attribute selection and discretization algorithm--Chi2. Classification   
   > clustering has reduced the two dimensional classification space to 13   
   > possibilities. Clearly my search space is still very big! I have had some   
   > success with population sizes of 5000.... but I tend to get stuck in local   
   > minimums a lot. Computational cost is huge! I have been looking into   
   > optimizing my population size and mutation rate. I have been playing with   
   > pop sizes of 1000-5000 and mutation rates of 0.01. My mutation operator   
   > randomly changes only one gene.   
   >   
   > Right now I don't know what I should do.....experiment with larger   
   mutation   
   > rates to keep diversity and prevent premature convergence into local   
   > minimums..... I have heard that small populations with larger mutation   
   rates   
   > can be more efficient than larger populations with smaller mutation rates.   
   > Shit- I worry that my endeavor is not feasible and that my search space is   
   > too damn big.   
   >   
   > Please help!!!!!!!   
   >   
   > the 'dude'   
      
   OK.... thanks for everyone's help (all 0 of you!). I take it that not many   
   people use this group- oh well. I worked things out by simplifying my   
   problem domain, which in turn reduced my features, classes, and search   
   space. By reducing my search space, my RCGA is able to adequately search the   
   solution space.   
      
   the 'dude'   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)