On sharpness of error bounds for multivariate neural network approximation
From MaRDI portal
Abstract: Single hidden layer feedforward neural networks can represent multivariate functions that are sums of ridge functions. These ridge functions are defined via an activation function and customizable weights. The paper deals with best non-linear approximation by such sums of ridge functions. Error bounds are presented in terms of moduli of smoothness. The main focus, however, is to prove that the bounds are best possible. To this end, counterexamples are constructed with a non-linear, quantitative extension of the uniform boundedness principle. They show sharpness with respect to Lipschitz classes for the logistic activation function and for certain piecewise polynomial activation functions. The paper is based on univariate results in (Goebbels, St.: On sharpness of error bounds for univariate approximation by single hidden layer feedforward neural networks. Results Math 75 (3), 2020, article 109, https://rdcu.be/b5mKH).
Recommendations
- On sharpness of error bounds for univariate approximation by single hidden layer feedforward neural networks
- Approximation by ridge functions and neural networks with a bounded number of neurons
- Error bounds for approximation with neural networks
- Ridge functions and applications in neural networks
- The ridge function representation of polynomials and an application to neural networks
Cites work
- A counterexample regarding ``New study on neural networks: the essential order of approximation
- A resonance principle with rates in connection with pointwise estimates for the approximation by interpolation processes
- Approximation by superpositions of a sigmoidal function
- Approximation results for neural network operators activated by sigmoidal functions
- Efficient estimation of neural weights by polynomial approximation
- scientific article; zbMATH DE number 3981746 (Why is no real title available?)
- scientific article; zbMATH DE number 3491650 (Why is no real title available?)
- scientific article; zbMATH DE number 3542691 (Why is no real title available?)
- scientific article; zbMATH DE number 495819 (Why is no real title available?)
- scientific article; zbMATH DE number 1405266 (Why is no real title available?)
- scientific article; zbMATH DE number 3191390 (Why is no real title available?)
- Multivariate Jackson-type inequality for a new type neural network approximation
- Multivariate neural network operators with sigmoidal activation functions
- Multivariate simultaneous approximation
- New study on neural networks: the essential order of approximation
- On best approximation by ridge functions
- On sharpness of error bounds for univariate approximation by single hidden layer feedforward neural networks
- On the degree of approximation by manifolds of finite pseudo-dimension
- On the near optimality of the stochastic approximation of smooth functions by neural networks
- Quantitative extensions of the uniform boundedness principle
- Rate of convergence of some multivariate neural network operators to the unit
- The errors of simultaneous approximation of multivariate functions by neural networks
- The sharpness of a pointwise error bound for the Fejér-Hermite interpolation process on sets of positive measure
- Universal approximation bounds for superpositions of a sigmoidal function
Cited in
(3)
This page was built for publication: On sharpness of error bounds for multivariate neural network approximation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6060396)
