XPost: comp.ai.neural-nets, sci.math.num-analysis, sci.math   
   From: ted.dunning@gmail.com   
      
   It doesn't really solve the problem, but support vector methods can   
   handle thousands of inputs with a feasible number of training examples.   
      
   Note, however, that this is dependent on the problem actually being a   
   low dimensional one that just happens to be phrased in high dimensional   
   terms.   In addition, the low dimensional nature of the problem has to   
   fit the assumptions of the method.   
      
   Bayesian methods can be essentially equivalent to SVM and thus can pull   
   the same sorts of tricks.   
      
   Essentially all of these combine a presumption about the simplicity of   
   the desired model with a measure of error.  The presumption of   
   simplicity is converted into a penalty for complex models and this is   
   used as a regularizer.  Bayesians think of this penalty as a prior   
   expectation, SVMers think of it as a performance bound on unseen data.   
   It works either way.   
      
   [ 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)