Arman Abdollahpour - Department of Computer Engineering, East Azarbaijan Science and Research Branch, Islamic AzadUniversity,Tabriz, Iran- 1Department of Computer Engineering, Tabriz Branch, Islamic Azad University, Tabriz, Iran
Mohammad Amin Karimi - Department of Computer Engineering, Tabriz Branch, Islamic Azad University, Tabriz, Iran
Siamak Ebrahimi - Department of Computer Engineering, Tabriz Branch, Islamic Azad University, Tabriz, Iran

People use various approaches for accurate diagnosis of the disease that one of the mostimportant methods is artificial neural networks (ANNs). Because to diagnose individuals’ diseasethe methods should be used that have a minimum of errors in diagnosis that ANNs have thischaracteristic. Therefore, this paper discusses the diagnoses of presence or absence of type 2diabetes by using ANN of Multilayer Perceptron (MLP). This paper tries to investigate thepresence or absence of type 2 diabetes using ANN of MLP with back-propagation algorithm. Inthis paper, we implemented ANN of MLP with back-propagation algorithm in MATLABenvironment to diagnose type 2 diabetes with data sets that we have which the performancecriterion is to maximize the accuracy of diagnosis of type 2 diabetes in the process of training andtesting. According to the Pima Indians Diabetes Dataset, the obtained result was that the trainingaccuracy was %0.78 and testing accuracy was %0.81.

Artificial neural networks (ANNs); Multilayer perceptron (MLP); Back-propagationalgorithm; Diabetes type 3; Pima Indians Diabetes Data Set; Accuracy of training; Accuracy oftesting