We are given n sampling times x1, …,xn and n corresponding targets y1,…,уn. For each i, we define a radial function: ф (x) = exp n Theregression model is defined as the function y(x, w) = wiФ1(x), where w = (wi,., wn) i=] is a vector of weights to be determined. We assume that the targets are related to the model by an additive noise e; as follows: yi = y(x1, w) +€i.
The noise samples are assumed to be independent and identically distributed as N(0,σ2).