Integral Representations in the Form of Neural Networks with Infinitely Many Units

Abstrakt

Model complexity of neural networks is investigated using tools from nonlinear approximation and integration theory. The estimates can be applied to a wide class of functions which can be represented as integrals with kernels corresponding to various types of computational units.

Kategorie a štítky

  • Konferenční příspěvek (mezinárodní konf.)