Newsgroups: comp.ai.fuzzy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!europa.eng.gtefsd.com!howland.reston.ans.net!EU.net!uunet!hasler!root
From: root@hasler.uucp
Subject: Re: Problem with Fuzzy Composition Recall
Message-ID: <CyrAFJ.68@hasler.uucp>
Organization: Dancing Horse Hill
In-Reply-To: fwatkins@ix.netcom.com's message of 1 Nov 1994 05:03:30 GMT
Date: Fri, 4 Nov 1994 18:38:06 GMT
References: <bchiu.783559351@balboa.eng.uci.edu> <394i72$nbh@ixnews1.ix.netcom.com>
Lines: 30


Brian Chiu writes: 

> ...
>	Given input vector X and corresponding output vector Y, the 
>fuzzy relational matrix R is then:
>					R = X o Y,
>where the "o" operation generates a matrix of the minimums between each
>X element and each Y element.  The Y is then recalled by doing:
>
>	Y = max(min(X,R)).
>
> ...
> ...  But I ran into a problem when the outputs Y are ALL larger than
>X, then, for example, X = [.3 .4 .7 .8]' and Y = [.9 .9 .9]', then
>the relational matrix R would be:
>				R = [.3 .3 .3
>				     .4 .4 .4
>				     .7 .7 .7
>				     .8 .8 .8 ];
>and the recalled output, max(min(X,R)) = [.8 .8 .8]', which is NOT Y!!
> ...
>	Is there a way around this problem ...

Can you prepend a one to X? ie, X = [.3 .4 .7 .8]' --> [1.0 .3 .4 .7 .8]'
-- 

John Hasler uunet!hasler!root
Dancing Horse Hill
Elmwood, WI 54740
