Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!travelers.mail.cornell.edu!news.kei.com!news.mathworks.com!newsfeed.internetmci.com!news.sprintlink.net!malgudi.oar.net!utnetw.utoledo.edu!news
From: dweddin@uoft02.utoledo.edu
Subject: BACKPROPAGATION "CONFIDANCE" MEASURES
Message-ID: <DE8nKH.GBp@utnetw.utoledo.edu>
Lines: 11
Sender: news@utnetw.utoledo.edu (News Manager)
Organization: University of Toledo
Date: Fri, 1 Sep 1995 09:06:08 GMT


Since there was some usion, I'll restate the question. By
"reliability measures" i was referring to a "confidence rating".
In other words, how "confidant" in the result is the 
backpropagation network.

If you toss out the output with low reliabilty ratings, the accuracy
goes up. This has already been done with RBF networks. I was 
wondering what work has been done on backpropagation. I remember
reading about it someplace.

