From: yaroslavvb@gmail.com   
      
   Suppose we consider a nested sequence of probability models (manifolds   
   of probability measures over A) with increasing number of degrees of   
   freedom m_0, m_1, m_2, etc.   
      
   How could I pick such a sequence such that the worst asymptotic bias is   
   as low as possible? For instance, if I have 0 degrees of freedom, and   
   looking at discrete probability distributions and measure bias in terms   
   of KL divergence, then it can be shown (ie Grunwald's "Game theory,   
   maximum entropy..") that the maximum entropy distribution minimizes the   
   bound on the worst asymptotic bias.   
      
   Does anyone know of any literature that uses this "uniform bias   
   reduction" idea to come up with model hierarchies?   
      
   [ 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)