home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.fuzzy
- Path: sparky!uunet!charon.amdahl.com!netcomsv!netcom.com!gperkins
- From: gperkins@netcom.com (Glen C. Perkins)
- Subject: Re: What does "AND" mean?
- Message-ID: <1993Jan22.170005.26421@netcom.com>
- Organization: Netcom - Online Communication Services (408 241-9760 guest)
- References: <1993Jan21.225423.25301@netcom.com> <PBM.93Jan21220053@mdis2.wdl.loral.com>
- Date: Fri, 22 Jan 1993 17:00:05 GMT
- Lines: 74
-
- pbm@mdis2.wdl.loral.com (Paul Melville) writes:
-
- >In article <1993Jan21.225423.25301@netcom.com> gperkins@netcom.com (Glen C. Perkins) writes:
-
- >> Suppose I'm looking for a wife "scientifically" (yeah, right) and I
- >> decide that my rule for evaluating candidates is "the prettier and the
- >> nicer, the more interested I am." Assume 0.0 PRETTY is elephant
- >> (wo)man, .5 is average, 1.0 is to die for, and 0.0 NICE is intolerable,
- >> .5 is average, 1.0 is an angel. If I decide to use MIN to represent AND
- >> (the most common definition of AND in fuzzy systems), I get:
-
- > [etc.]
-
- >A thought: You seem to imply that your membership rules for both PRETTY
- >and NICE are linear. i.e. if Miss X is "twice" as pretty as Miss Y,
- >then,
-
- > mPRETTY(X) == 2 * mPRETTY(Y)
-
- >what about first choosing a more realistic membership function, one
- >that reflects how you feel about the relative importance of that LEVEL
- >of PRETTY-ness or NICE-ness. Like this:
-
- >mNICE(x)^
- > 1.0| *****
- > | ********
- > | ******
- > | ****
- > | ***
- > 0.5| **
- > | *
- > | *
- > | *
- > | *
- > 0.0+---------------------------------------> x
- > 0 1 2 3 4 5 6 7 8 9 10
- > ^ ^ ^
- > | | |
- > downright ave. angel
- > nasty
-
- >where "x" represents your linear rating of the marital candidate.
- >I still think that you are right about wanting soe kind of anD beside
- >MIN, but if you calculate membership in the two sets with some kind of
- >(easily defined) curve similar to above, then a simple average ight be
- >OK, or RMS or whatever.
-
-
- I'm glad I asked, because this is the kind of suggestion I was
- looking for. I actually wasn't using linear membership functions,
- but in my consideration of the appropriate rules I was
- overlooking the information already encoded in the mem function
- and sort of unconsciously assuming something simple and linear.
- Using a linear rating system and then running it thru a
- PRETTY_ENOUGH and NICE_ENOUGH mem funtion and then ANDing those
- makes a LOT more sense. Thanks.
-
- I'm still interested in other good suggestions such as what
- AND functions would be most reasonable. This sort of thing has
- probably been done somewhere in some form. When I say "OR"
- in conversation, I really do think in terms of a MAX function,
- unconsciously of course. At least, I sense that the MAX
- function fits my "feeling" of what I mean when I casually
- say "or". I'm much less confident of AND, and it doesn't
- really matter except when I decide to generate fuzzy rules
- to model human psychology (in some narrow domain.)
- --
- +-----------------------+---------------------------------------------------+
- |--- Glen C. Perkins ---| |
- | <gperkins@netcom.com> | "Don't forget, your mind |
- | Native Guide Software | only *simulates* logic." |
- | Palo Alto, California | |
- +-----------------------+---------------------------------------------------+
-
-