I wanna ask about Adaptive Neuro Fuzzy Inference System (ANFIS) method. I used clustering to find fuzzy rules and used the mean and deviation standard of each cluster for initialization of the first premise parameters (c and a) of membership function. c is from the mean, and a is from the deviation standard

Now, my problem is I find that there's a cluster that have deviation standard = 0 (zero), so it can't be premise parameter (where the deviation standard will be "a" parameter of membership function) because it will the gbell membership function undefined.

If a=0, how can I calculated the gbell membership function of ANFIS? Is there any solution of finding rules and membership function of ANFIS that could be implemented in clustering? Because I have to make ANFIS manual and I was banned to use MATLAB and I should make ANFIS in java programming and I have to do the manual first.

Thank you.

  • $\begingroup$ A cluster has zero standard deviation iff it is constant. $\endgroup$ Dec 31, 2016 at 17:53
  • 1
    $\begingroup$ Your question is rather hard to understand. Can you try to clarify? $\endgroup$ Dec 31, 2016 at 17:54
  • $\begingroup$ @Yuval Filmus. So, I mean that I will use the standard deviation for the parameter "a" of gbell membership function. But if a=0, gbell membership function can't be calculated. $\endgroup$ Jan 1, 2017 at 5:54
  • $\begingroup$ I have revised it $\endgroup$ Jan 2, 2017 at 10:32


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.