What does overlapping m.f.'s really represent?

Ulf Nordlund (ulf.nordlund@pal.uu.se)
Mon, 29 Jan 1996 18:11:34 +0100


What exactly does the "density" in a resulting fuzzy set represent?

By "density" I mean the amount of "overlapping" of contributing fuzzy
sets. We normally construct our resulting fuzzy sets by OR-ing together
the sets derived from idividual rules. In doing so, we do not give any
regard to the fact that several sets may be referring to the same area of
the membership function - i.e. they overlap. Although I can understand the
reasons for doing it this way, I can also see some reasons for taking the
"density" into account as well. (High density means that many of the rules
- many fragments of stored "knowledge" - point towards the same thing.
Note that the "non-overlapping" rules do not have to be wrong because of
this.)

Comments?
References?

Thanks,
Ulf