XPost: comp.ai.neural-nets, comp.databases, comp.ai.fuzzy   
   XPost: sci.math   
   From: ted.dunning@gmail.com   
      
   The suggestion to hand-write rules to define abnormal situations pretty   
   much describes the state of the in fraud detection 15 years ago.   
      
   Adaptive techniques, notably including anomaly detection for the first   
   few generations completely destroyed hand-written systems, especially   
   in credit card fraud.  The problem isn't so much that the hand-written   
   systems don't work initially, but that they decay over time.  My term   
   for the phenomenon is bit-rot.  As new rules are added and as the world   
   changes, the hand-written rules systems eventually produce really,   
   really bad results.   
      
   That said, rules do have their place.  The feature detectors that go   
   into adaptive systems are definitely rules of a sort (but they don't   
   themselves predict fraud very well) and the business logic that   
   determines what actions to take are also rules.  Rules match a   
   regulatory environment (as in, don't call a customer about suspected   
   fraud more than once in any 90 day period).  Adaptive learning methods   
   work much better at balancing the risk of error and the value of   
   detection.   
      
   [ 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)