From: voisin.bruno@wanadoo.fr   
      
   Hi,   
      
   I am used to neural nets, and their statistical validation, i.e. keeping   
   colinearities low, using cross-validation etc...   
   I have coded a FL library (c++) and I can now solve simple to intermediate   
   FL problems using interconnected FL controllers.  The application will   
   mostly be data modelling, but fuzzy sets and rule sets may not be stable   
   over time, and some fuzzy variables may even become simply redundant.  For   
   stability, I would also assume the overall FL system architecture must be   
   kept as simple as possible.   
      
   I would like to know FL practitioners deal with this?   
      
      
   Many thanks in advance,   
   Bruno   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)