Initializing fuzzy c-means

Bob Dalton (red@swl.msd.ray.com)
Sun, 26 Apr 1998 18:02:10 +0200 (MET DST)

Dear Jonathan: Some time ago, I think 1993-94, Ron Yager published
a note in Systems, Man and Cybernetics on something called
the "Mountain Method", which could be used to initialize
memberships in the c-means algorithm. Although Yager
suggested his method might speed convergence of c-means,
he didn't prove it or furnish experimental data supporting
his claim.

In my experience, using a crisp initialization that satisfies
the constraints works as well as any other technique. I'd
be VERY interested if you found something better.

Bob Dalton
--------------------------------------------------------------------
--------------------------------------------------------------------

Robert E. Dalton, Ph. D. email: red@swl.msd.ray.com

Raytheon Co. Mailstop: T3MN35 voice: (978) 858-5927

50 Apple Hill Drive, Tewksbury, MA 01876

--------------------------------------------------------------------
--------------------------------------------------------------------