Newsgroups: comp.ai.fuzzy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!lerc.nasa.gov!magnus.acs.ohio-state.edu!math.ohio-state.edu!howland.reston.ans.net!swrinde!hookup!news.mathworks.com!zombie.ncsc.mil!nexus.coast.net!harbinger.cc.monash.edu.au!newshost.anu.edu.au!csis!mac
From: mac@cbr.dit.csiro.au (Mark Cameron)
Subject: Fuzzy certainty measures?
Message-ID: <DAM3rF.8vL@cbr.dit.csiro.au>
Organization: CSIRO Division of Information Technology, Canberra, Australia
Date: Fri, 23 Jun 1995 06:09:06 GMT
Lines: 32

I have a question about confidence distributions over a fuzzy output variable.
The scenario is this:
	Several fuzzy systems each provide an output variable described with
	some membership function over the same output domain.

	The membership functions completely match over some subdomains, and
	don't match to a greater or less extent over the remaining subdomains
	of the output domain.

The questions are:
	Are there tools/techniques for correlating the amount of 'confirmation'
	between the output variables? If all systems 'agree' that the membership
	function of the output variable takes some particular shape over
	some subdomain, I would like to attatch more 'confidence' to that 
	section of membership. Conversely, the less they 'agree', the less
	'confidence' I have in the membership function.

	I don't want to 'merge' the separate systems into a single system
	unless I absolutely have to. (You can view each system as an 'expert'
	in some domain, where the output fuzzy variable represents common
	expertise or common understanding).

If there is an article that addresses this, let me know. If you think
it's krazy, tell me why! If the measure has a name, please share your knowledge.
I look forward to your responses.

Mark.
-- 

mark.cameron@dit.csiro.au
CSIRO Division of Information Technology
GPO Box 664,                                         Tel: +61 6 216 7035
