Optimization of Neuro-Fuzzy System
Moechammad Sarosa, Akhlaque Ahmad, B. Riyanto, Achmad Saefuddin Noer · ITB Journal of Information and Communication Technology · 2007
Neuro-fuzzy system has been shown to provide a good performance on chromosome classification but does not offer a simple method to obtain the accurate parameter values required to yield the best recognition rate.This paper presents a neuro-fuzzy system where its parameters can be automatically adjusted using genetic algorithms.The approach combines the advantages of fuzzy logic theory, neural networks, and genetic algorithms.The structure consists of a four layer feed-forward neural network that uses a GBell membership function as the output function.The proposed methodology has been applied and tested on banded chromosome classification from the Copenhagen Chromosome Database.Simulation result showed that the proposed neuro-fuzzy system optimized by genetic algorithms offers advantages in setting the parameter values, improves the recognition rate significantly and decreases the training/testing time which makes genetic neuro-fuzzy system suitable for chromosome classification.