In a neural network field, there are several techniques how data can be pre-processed (normalization, standardization, whitening...) in order to obtain higher accuracy of a network without changing its architecture. Even if fuzzy is broadly spread in data/image processing, I did not find any pre-processing technique based on fuzzy. There are fuzzy neural networks, but they do not realize data pre-processing without original architecture modification. Also, a transformation from the original data range into [0, 1] cannot be considered as a fuzzy pre-processing.
So, the question is: do you know some fuzzy-based data pre-processing technique?
Dear Petr,
I suggest you to see links and attached files in topic.
- Fuzzy Neural Network Modeling for Product ... - Science Direct
https://www.sciencedirect.com/science/article/pii/.../pdf?...
- Data mining using dynamically constructed recurrent fuzzy neural ...
www.ntu.edu.sg/home/elpwang/PDF_web/98_PAKDD.pdf
- Neural Fuzzy Systems
uni-obuda.hu/users/fuller.robert/ln1.pdf
-Comparing a Neural-Fuzzy Scheme with a Probabilistic Neural ...
https://pdfs.semanticscholar.org/.../f202268883cc5f92250e5b5ca2...
- Modular Neural Network Preprocessing Procedure with Intuitionistic ...
https://link.springer.com/chapter/10.1007/978-3-319-26154-6_14
Best regards
@Mohamed-Mourad Lafifi Thank you for the repply, but none of the papers describe fuzzy pre-processing which can be applied to a general neural network architecture. This is exactly what I write in my question: "There are fuzzy neural networks, but they do not realize data pre-processing without original architecture modification."
- Badges
- Science topic