Eskişehir Türk Dünyası Uygulama ve Araştırma Merkezi Bilişim Dergisi

Yazarlar: Serel ÖZMEN-AKYOL, Eyyüp GÜLBANDILAR

Konular:-

Anahtar Kelimeler:Artificial Neural Networks,Fuzzy Neural Networks,Adaptive-Network Based Fuzzy Inference Systems,Pattern Classification

Özet: Pattern classification is also widely used in industry and medicine. While the classical observational classification is made, the errors in classification due to personal mistakes occur. Artificial learning methods are usually used to reduce these classification errors. In this study, it is aimed to classify iris flower species by using the artificial neural network (ANN) and fuzzy artificial neural network (ANFIS). In this study, it was preferred to use the sample data of iris flower widely used. The selected ANN model has three layers, feedforward, four inputs and one output. In the ANN model training was used the 90 sample data and in the testing process were used 60 sample data. The selected ANFIS model has two fuzzy sets for each input neuron, four inputs, and one output. In this model, training and testing were carried out using similar samples. For both models, the actual values and predicted values were compared. It was obtained low error rates. The classification results of ANN and ANFIS methods were compared and it was found that ANN method produces closer estimations than ANFIS method.  

---