From: ted.dunning@gmail.com   
      
   On May 7, 2:29 am, "A.L."  wrote:   
   > On Sun, 06 May 2007 13:18:47 GMT, "TiGeR"  wrote:   
   > >Hello.   
   >   
   > >I've got a problem.   
   >   
   > >During a test of back propagation algorithm I wanted to teach   
   > >the net the sin(x*y) function with a domain of [-2;2] for   
   > >both variables.   
   >   
   > >But it turned out that it can't be done.   
      
      
   I concur that you have a problem and that your correct course of   
   action is to fix the bugs in your program.   
      
   3-4 neurons should suffice to get a very rough approximation on any   
   sub-interval of the range -pi, pi.  Adding neurons should allow you to   
   get progressively more refined approximations.   
      
   A common mistake is to not include a constant input.   
      
   There are standard sets of problems with known answers that you might   
   test your software on.   
      
   [ comp.ai is moderated ... your article may take a while to appear. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)