şengöz, nilgün mehmet akif ersoy üniversitesi - stratejik işbirliği proje danışmanlık eğitim uygulama ve araştırma merkezi, Burdur, turkey , özdemir, gültekin süleyman demirel üniversitesi - mühendislik fakültesi - endüstri mühendisliği bölümü, Isparta, Turkey

Comparison of ANFIS and Cascade Correlation Neural Network Methods for Classification Problems

25574

In this study, it focused on classification problems. This thesis differentstructure with 3 data set (Seed, Terrain Satellite Imaging and Red Wine Quality)examination and the same algorithms applied to these data as the relationship aswell as the performance achievements and time in terms between the eyes reveals.Seed data is for the comparison with the ANFIS and CCNN test performance of86.41% and 88.06% percent, in the classification process has always CCNN lesstime when viewed from the side. In the dataset Terrain Satellite Imagingperformance ANFIS percentage of 100%, in the CCNN is 99.92%. The duration ofthe transactions made in the algorithm s data ANFIS 800 seconds, it completed theclassification process CCNN all in 72 seconds. Our last data set is red wine qualityevaluation and performance when viewed as ANFIS% 99 975, the CCNN is clearlynot constitute a very significant difference between each other like 99 862%.According to the terms of processing time with the BKS method ANFIS 85, 271seconds is far too cumbersome.

125

Artifical intelligence , Neural network , ANFIS , Cascade correlation neural network , CCNN

Journal Of Natural an‎d Applied Sciences

130