home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.fuzzy
- Path: sparky!uunet!gatech!usenet.ins.cwru.edu!agate!stanford.edu!kronos.arc.nasa.gov!butch!netcomsv!netcom.com!gperkins
- From: gperkins@netcom.com (Glen C. Perkins)
- Subject: What does "AND" mean?
- Message-ID: <1993Jan21.225423.25301@netcom.com>
- Organization: Netcom - Online Communication Services (408 241-9760 guest)
- Date: Thu, 21 Jan 1993 22:54:23 GMT
- Lines: 91
-
- I've been wondering the meaning (one meaning) of a fuzzy AND as I've
- been attempting to put together some fuzzy rules for a specific
- psychological application. My actual application is a little obscure,
- but I have an appropriate analogy that demonstrates my confusion
- perfectly:
-
- 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:
-
- .7 PRETTY AND .5 NICE = .5 INTERESTED
-
- Okay, fine, but this means that she could get prettier approaching
- 1.0 ("to die for") without my interest increasing at all. In other
- words, if she were .5 NICE, I WOULDN'T EVEN CARE whether she was
- average (.5) or gorgeous (1.0), it would be just the same to me.
- Needless to say, that's not what I observe! At least in this case,
- when I say "AND" I don't seem to mean MIN.
-
- Well, what DO I mean? I don't know. Any increase in either NICE OR
- PRETTY seems to increase my interest, so when I say AND do I really
- mean OR? Well, not if OR means MAX because that would yield
- something like "as long as she's PRETTY, it doesn't matter how NICE
- she is." Wrong again. Both matter, so I don't want either PRETTY or
- NICE to cut off input from the other.
-
- If I multiplied the two values (algebraic product), it would work
- better, but it doesn't make sense that average (0.5) PRETTY AND
- average (0.5) NICE = significantly below average (0.25) interest.
- (There are other fancier products: an Einstein product would yield
- 0.2, and a Hammacher product would yield 0.33, but so what?) And if
- multiplication is AND, what would make a complementary OR?
- Addition? Probably, but how do you normalize it so that two
- variables in the range [0,1] add together to form a variable in the
- same range? If you just divide by two, that works fine, but it makes
- OR an average. Is that what OR "should" be?
-
- Maybe my conversational use of AND really implies two independent
- criteria which should have separate rules instead of one rule with an
- AND in it. One rule might say "the prettier, the more interested I am"
- and a second rule might say "the nicer, the more interested I am."
- Well then you'd still have to combine the output of the two rules by
- some means, typically a centroid which in this case is just the
- average again. Is that what AND "should" be?
-
- I guess we could fold them back into one rule and define AND as an
- average instead of MIN. MIN is the most common interpretation of
- AND, but not the only one. Defining fuzzy AND this way, I'd really be
- saying "my interest in her is proportional to the average of PRETTY
- and NICE."
-
- Well, that's not really right either. I DO have SOME standards! If she's
- intolerably nasty, I don't care if she's drop dead gorgeous, I'm not
- interested. Or... do I mean that my interest is just VERY low? I'm not
- totally sure myself, but I AM sure that my interest is not as high as
- average (0.5). Maybe my AND really is somewhere between MIN and
- AVERAGE, a linear combination of MIN and AVG. As I understand
- some types of decision theory (including fuzzy logic) this is not
- unheard of, but still uncommon.
-
- Ultimately, it appears that the only way to model this deceptively
- simple and I think typical psychological rule: "the prettier and the
- nicer, the more interesting" would be for me to sit down and build a
- truth table. I'd probably need to make one axis PRETTY and one axis
- NICE in, say, increments of 0.1, and then ask myself how I'd feel
- about each combination. Then I'd try to figure out what the function,
- INTEREST(PRETTY, NICE) really was.
-
- This sort of seems like I'm missing the benefits of fuzzy logic's
- "ease of development." If I have to design a fancy algorithm for the
- AND in my one rule, why not just throw away the rule and just use
- the algorithm? I know fuzzy logic isn't the answer to every question,
- but it does seem as though it ought to be ideal for encoding
- something like my little rule.
-
- Am I missing something? (My marbles, perhaps?) I'd be interested in
- hearing a bit of a discussion if anyone is interested, so don't e-mail
- me personally (unless you want to introduce me to a (1.0, 1.0). I
- suppose that would be the '90s equivalent of a "10"!)
-
- --
- +-----------------------+---------------------------------------------------+
- |--- Glen C. Perkins ---| |
- | <gperkins@netcom.com> | "Don't forget, your mind |
- | Native Guide Software | only *simulates* logic." |
- | Palo Alto, California | |
- +-----------------------+---------------------------------------------------+
-
-