optimizing fuzzy logic membership functions with genetic algorith

hello everyone, I want to ask for help for my project. My project is about optimizing fuzzy logic membership functions with genetic algorithms. Can anyone help me? thank you

[Ameer Hamza]

Can you further elaborate on your question? What type of objective function do you have?

[Dicky Gultom]

Thank you for replying to my question.

I have a project to optimize the member function of fuzzy logic with the Genetic algorithm. The case with scholarship data. In this scholarship data has variables namely: Salary, value, and attitude points. So my project is to optimize the limits of fuzzy logic for these variables. For example salaries will consist of 3 categories of fuzzy sets that are high, medium, low. Now to make these limits for these values. Examples of high salaries range from 5 million to 10 million, maybe. Like that. thanks.

[Ameer Hamza]

If you are able to write an objective function, which takes the limits of the fuzzy variables as input and outputs the goodness of those limits (small values if input limits are good and a large value if input limits are not suitable). If writing such a function is possible, I don't see any reason why ga() from global optimization toolbox cannot be used to optimize those limits.

[Dicky Gultom]

true the problem is I want to make 3 areas in the fuzzy set for example before the salary consists of 3 regions, namely high, medium and low but can not create limits in the form of the figure. So use GA, right?

[Ameer Hamza]

GA is an optimization algorithm. It does not care about which function you are trying to optimize. It just treats your objective function as a black box. It inputs some values, and get the output, and based on those values. It tries to search the optimal point.

So, the first problem is to write a function that works like this: I input it some random limits for the fuzzy variables. It needs to tell me if those limits are "good" or "bad". You can use whatever logic to decide whether the input limits are good or bad. The best logic is the one which correctly captures your definition of "good limits"

[Dicky Gultom]

Thank you for your input. Means I have to make a function to make these restrictions so that GA can optimize it? Is it like that ?

[Ameer Hamza]

yes, the first step is to define a function, which can decide if the given limits of membership function are good or bad. Say, I give it input that the first membership function is defined on the interval [0, 2], second is defined on the interval [1 7], and the third one on the interval [5 10]. It needs to process your dataset according to these limits and output (as a number) if these limits are good or bad for your final goal.

[Dicky Gultom]

sorry I want to ask who made the function and the interval is it made by us based on the output we have. where this output is our target?

[luis lara]

Sorry, you can do it, I have the same question about how to optimize the membership functions?