• CONVEX APPROXIMATION BY A FEED-FORWARD NEURAL NETWORK
Abstract
In this paper, we study a feed-forward neural network with one hidden layer of constant units and a linear output and we can approximate function in Lp,r arbitrarily as long as the number of hidden nodes is sufficiently large. Then we give the formula for the upper bound of the approximation error in Lp,r norm by the constructive a convex feed-forward neural networks. This result is used to formulate a new method of constructing neural network for approximating a measurable function with Lp,r norm.
Keywords
Weighted spaces, convex approximation, feed-forward neural network, steklov mean.
Full Text:
PDF
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
| © 2010-2022 International Journal of Mathematical Archive (IJMA) Copyright Agreement & Authorship Responsibility |