بناء نموذج شبكة عصبونية باستخدام التحليل العاملي و خوارزمية K-Means

الملخص EN

This paper aims to propose a multilayer neural network model based on factor analysis and k-means algorithm by integrating them into the neural network architecture.

factor analysis reduces the number of primary data dimensions and puts the output factors into the input layer.

the data is then collected according to k-means algorithm depending on extracted factors to obtain two clusters representing the decision variable by engaging all relevant variables in the algorithm.

hereafter, the output is placed in the output layer; then the proposed neural network model is trained.

for the purpose of comparison, the model was tested on a database of multivariat for the electronic library at the faculty of economics at the university of aleppo.

the proposed model showed a decrease in complexity degree through a decrease in the number of weights.

in addition, it showed a high classification accuracy and a lower training time than an unmodified ANN model.