From: wallace@image.ntua.gr   
      
   >   
   >   
   >i'm looking into a clustering algorithm (modififcation of standard k-means   
   >to start with) for fuzzy data.   
   >   
   >my question is this: are there existing standard distance measures between   
   >data instance of this type?   
   >   
   This is a very interesting question that could be the beginning of a   
   very stimulating exchange of opinions on the thread :-)   
      
   More or less, this is how things go: You take a standard crisp algorithm   
   and decide to extend it to the fuzzy case. Fortunatelly, this is not a   
   one-to-one mapping. Depending on the fusyy metrics and measures you   
   deside to utilise in order to extend crisp distances and measures, you   
   have a different implementation. My experience shows that there is no   
   right and wrong selection. Depending on the data set and the wished   
   result, the selection may differ a lot.   
      
   In other words, in the crisp case the selection of the algorithm is the   
   most important step. But when moving to the fuzzy case, all the interest   
   moves to the selection of the operators and measures. So, bottom line   
   is: I am afraid no one will be able to help you too much; the measures   
   you are looking for are for you to deside / select depending on your   
   application.   
      
   I sincerely hope I helped clarify some things rather than confused you   
   some more,   
   Manolis   
      
   --   
   Manolis Wallace   
   http://www.image.ece.ntua.gr/~wallace/   
   Image, Video & Multimedia Systems Laboratory   
   Department of Computer Science   
   School of Electrical & Computer Engineering   
   National Technical University of Athens   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)