Difference between Testing and Checking Data in Fuzzy Logic Toolbox - Neuro-Fuzzy Designer

Can anybody clarify the difference between 'testing data' and 'checking data' in the Neuro-Fuzzy Designer? I understand the purpose of testing a trained system against a unique equivalent dataset. However, the Neuro-Fuzzy Designer appears to make a distinction between 'testing' and 'checking'.

As far as I can see, the user manual seems to use the two terms interchangeably. Could anybody clarify the functional difference between the two? Apologies if I am missing something obvious.

[Greg Heath]

No apologies needed. I've been designing NNs for more than 29 yrs and I have never encountered the term in that context!

My concept of "checking" is

1. Periodically review/check your work while you are doing it

2. Make a final check once you have finished it.

Hope this helps.

Greg

Answer 1

[Sudheer Nuggehalli]

If you navigate to the "Model Validations Using Testing and Checking Data Sets" section in the documentation link below, there is an explanation on the differences between the Testing and Checking data sets, as well as the model validation process in general. Hopefully this helps in clarifying the difference.

Neuro-Adaptive Learning and ANFIS