How to use cross-validation with ANFIS in fuzzyLogicDesigner of MATLAB?

I'm looking to build a rule-based fuzzy system using adaptive neural networks (ANFIS) through Fuzzy logic Toolbox. However, since I don't have a large amount of data, I would like to use the cross-validation technique, but when I click on "Tuning," this option is not available when using ANFIS as the optimization method.

How should I proceed? Is there any tool available in MATLAB that allows me to use Cross-Validation within the fuzzy toolbox with ANFIS?"

Thanks

Answer 1

[Arka]

Hi Maria,

From what I understand, you want to build a rule-based fuzzy system using “ANFIS” via the “Fuzzy Logic Toolbox”, but you are not getting the option to use the cross-validation technique when you click on “Tuning”.

The option to use cross-validation for tuning “ANFIS” models is not directly available in the “Fuzzy Logic Toolbox”, but you can manually implement cross-validation by partitioning your data into training and testing subsets, and then training and evaluating your model.

Here is an example of how you can do the same:

1. Split the data into training, cross-validation, and/or testing subsets. This can be done manually or by using functions like “cvpartition”.
2. Create the ANFIS model using the “anfis” function, using the training data.
3. Evaluate the model using the testing data using the metrics that are of interest to you.
4. Repeat steps 2 and 3 for cross-validation set.
5. Average the metrics across all sets to obtain the assessment of the ANFIS model’s performance.

Apart from this, you can also use other MATLAB toolboxes like “Statistics and Machine Learning Toolbox” or the “Neural Network Toolbox”. They provide more comprehensive support for cross-validation and other optimization methods.

To learn more about “cvpartition” and “anfis”, please go through the MathWorks documentation links below:

1. https://www.mathworks.com/help/stats/cvpartition.html
2. https://www.mathworks.com/help/fuzzy/anfis.html

Thank you.