On the interface we present the

approximation error in the form

and the (2N + 1)-term

approximation error is given by

Our estimates improve the previous known results in several aspects: Firstly, the anisotropic

approximation error is obtained in a different way.

j]) basis functions and it is shown that under sufficient smoothness an

approximation error of order [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] can be achieved.

However, in both presentations one can only find fairly crude estimates for the related global

approximation error.

2 is the

approximation error and the boundary term involved there is specifically generated by the discretization of variational inequalities.

Based on the weighted-residual error estimator from [7], we introduced an overall error estimator which controls both, the discretization error as well as the data

approximation error (Theorem 3.

J]) of symbols of the multiple Gabor multiplier that minimizes the

approximation error [[parallel]T - [G.

3, the smallest

approximation error was achieved for Learning rate: 0,01, Momentum: 0,4, Sigmoid's alpha value: 0,4, Neurons in first layer: 2.

It has been noticed that, the interrelation between the growth of an entire function in terms of [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] and

approximation error in [L.

Q] approximating this error, and random

approximation error [E.

The procedure is halted when the desired number of the linear segments M is reached, or the

approximation error is below given threshold [epsilon].