New study on neural networks: the essential order of approximation.

Abstract

For the nearly exponential type of feedforward neural networks (neFNNs), the essential order of their approximation is revealed. It is proven that for any continuous function defined on a compact set of R(d), there exist three layers of neFNNs with the fixed number of hidden neurons that attain the essential order. Under certain assumption on the neFNNs, the ideal upper bound and lower bound estimations on approximation precision of the neFNNs are provided. The obtained results not only characterize the intrinsic property of approximation of the neFNNs, but also proclaim the implicit relationship between the precision (speed) and the number of hidden neurons of the neFNNs.