I got a problem in how to deside the initial parameters of fuzzy membership functions.
in premise or consequent universe.
The type of the memberships is bell function or gause function that composed
by three parameters, they are center, standard deviation, and slope.
My question is how to decide the initial parameters before learning or inference in
order to get a faster or better result, it means we can take less time to finish the
job and get less RMSE(Root Mean Square Error) with the optimal initial parameters.
How to decide them. Please give me some sugestions.
Thanks a lot.
My E-Mail is : ee837314@eesun1.dyit.edu.tw
FAX : (04)8531904
Shinn-Wen