“… where M is the total number of training patterns. Now, the input–output mapping function of the normalised RBF network takes the following form where d j is the linear weight applied to the basis function [15], and the denominator term represents the Parzen–Rosenblatt density estimator, which consists of the sum of ‘ M ’ multivariate Gaussian distribution centred on the data point p 1 , p 2 , p 3 , …, p M . In (23), the centres of the normalised RBF coincide with the data points .…”