how to solve the classification problem with anfis?

Generally, the anfis is used to solve the regression problem.If there is no model provided,the anfis will use the given data to generate the inference system and adjust the parameters in order to get the best model. I want to solve the classification problem by anfis, and tune parameter to get optimal recognition rate.Thereby if i choose to generate one grad function for every attribute and there are three classes,it will be three rules,but the current anfis process the whole data rather than partial data to generate rulebase and warn that "more than one rule".What's more,after browsing the anfis code,I understand the modeling process for regression,but i didn't find the code for data calculating. In order to improve the recognition rate,i want to deal with the data in the class by fcm, if there are two clusters in every class, there will be six rules for dataset with three classes.I try to solve the problem,unfortunately i failed. Maybe i didn't express my problem clearly, simply,how to solve the classification problem(such as the classical iris dataset) with anfis? I will be very appreciate if you can guide me or provide me with some helpful articles.

Answer 1

[usman ashraf]

I am having same problem. can you please guide if you find the solution?