| In this presentation we look on symmetric, positive definite functions from the
point of view of the frequency domain. In the machine learning community these functions are
referred to as kernels. Taking a spectral approach is quite instructive. First of all, translational and
rotational invariant kernels can adequately be characterized by their spectral representation, and
secondly, the dual concept of a regularization operator by means of an induced Green’s function
becomes more evident and handy. |