[A method for creating fuzzy neural-network models using the MATLAB package for biomedical applications]

Area of Science:

• Biomedical Engineering
• Computer Science
• Artificial Intelligence

Background:

• MATLAB-based simulations of hybrid neural networks present challenges in biomedical data classification.
• Existing methods lack integrated approaches for constructing and tuning fuzzy neural network models.

Purpose of the Study:

• To address the problems of simulating hybrid neural networks for biomedical data classification in MATLAB.
• To propose graphic interfaces and algorithms for constructing and merging fuzzy neural network models.

Main Methods:

• Utilizing neural network structures as defuzzifiers within fuzzy systems.
• Developing graphic interfaces for building fuzzy neural network models.
• Implementing an algorithm for fuzzy neural network model construction, including membership function tuning and perceptron-based defuzzifier parameter tuning.

Main Results:

• Suggested graphic interfaces facilitate the construction of fuzzy neural-network models.
• Methods for merging models from different MATLAB packages enable integrated analysis.
• The proposed algorithm effectively tunes fuzzy system fuzzifier membership functions and defuzzifier parameters.

Conclusions:

• The developed approach enhances the simulation of hybrid neural networks for biomedical data classification.
• The proposed methods and algorithm offer a robust framework for building and optimizing fuzzy neural network models in MATLAB.
• This work contributes to advancing AI applications in biomedical diagnostics through improved simulation techniques.