Neural Network Ensemble Based on K-Means Clustering Individual Selection and Application for Software Reliability Prediction

A novel neural network ensemble is proposed and applied to the software reliability prediction in the paper which based on the K-means clustering individual selection. First, multiple neural networks are generated by changing the structure of the neural network, then individual selection ensemble is made with K-means clustering method, and finally the outputs of these selected individuals by entropy weight method are integrated. The new method has been proved superior in software reliability prediction by experimental comparison.