Navigation(slides)

Section 6.3: Radial Basis Function Networks

Radial basis functions depend on the radial distance from the center, latex($\bar{\mu}_{j}$) such that

$\phi_{j}(\bar{x}) = h(|| \bar{x} - \bar{\mu}_{j}||)$

They were originally used to produce a function that could exactly interpolate between a set of points. This was done by equation *6.38* The coefficients were found by least squares. This function produced an over-fitted function that fit every training value exactly.

Classes/BMTRY790/KernelMethods/010_Radial_Basis_Function_Networks (last edited 2008-01-29 13:20:19 by strasbu)