The Errors of Approximation for a Class of Approximate Interpolation Neural Networks in the L~p Metric
Lu Wenxiu · Journal of Xianyang Normal University · 2013
For a class of approximation interpolation feedforward neural networks(ai-nets) with single hidden layer,the problem of approximation is studied in this paper.With the Steklov mean function,the errors for the interpolation neural networks approximating Lebesgue integrable functions are estimated by the modulus of smoothness.It is shown that for a ai-net can approximate,with arbitrary precision,any p-th(1≤ p +∞)Lebesgue integrable function f(x) defined on [a,b] as long as the number of hidden nodes n is suffciently large.