Two questions about Fuzzy Logic

Hi,

I have two questions about Fuzzy Logic that I would be grateful is somebody could answer.

1) Can the outputs of a Fuzzy Logic system be specific values, for example "Yes" or "No". Take for example the Tipping Demo in the documentation. The value of the Tip is based on the quality of the service and the food. Could this be modelled as - give a Tip, "Yes" or "No", based on the quality of the service and the food?

2) In the anfis GUI,it forces you to specify the rules before you can train the network. But what if you do not know the rules governing the system. For example in the Tipping Demo we know the rules before hand, but in other applications we will not necessarily know the rules. To me the whole point of training is to determine the rules and "teach the network". Am I interpreting this correctly?

Appreciate any comments

Thanks

John

[Merged information from duplicate Question]

Hi,

I am reading through the Fuzzy Logic "Tipping" Demo in the documentation, available here http://www.mathworks.co.uk/help/toolbox/fuzzy/fp130.html#FP207

In the Demo, the value of the Tip is based on the quality of the service and quality of the food.

My question is, could this be modelled instead as the decision to give a Tip, "Yes" or "No", based on the quality of the service and the food?

How can you model crisp "Yes" or "No"outputs? How would the shape of membership function change?

I want to practice training a neuro fuzzy logic system using this example. So I generated some training data for quality of food , quality of service and output decision Tip "Yes"or "No". Do I need to change the membership functions?

I cannot find a similar example elsewhere.

Thank for the help

John

Thank you

I

Answer 1

[Walter Roberson]

Your existing Question is still here and still active. Please do not open duplicate Questions. You can edit your existing Question to provide more information, and you can post comments to remind people about your question.

Your duplicate Question has been removed.

[John]

Hello,

Would anybody be in a position to offer advice on this question?

Thank you

John

[Walter Roberson]

There was a question similar to your #2 a few weeks ago, that an answer was given for; unfortunately I was not able to locate it when I searched.

I do not know for your #1. A common method for a clear binary output is to threshold the output condition, such as OUTPUT >= 0.5 . I do not know if that can be embedded as a FIS rule (the one time I examined FIS rules, the logic escaped me.)