Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks

Neural Networks

Volumn 3, Issue 5, 1990, Pages 551-560

  Hornik, Kurt ^a   Stinchcombe, Maxwell ^a   White, Halbert ^a  

^a UNIVERSITY OF CALIFORNIA   (United States)

Author keywords

Approximation; Derivatives; Feedforward networks; Sobolev space

Indexed keywords

MATHEMATICAL TECHNIQUES--APPROXIMATION THEORY;

FEEDFORWARD NETWORKS; HIDDEN LAYER ACTIVATION FUNCTION; MULTILAYER FEEDFORWARD NETWORKS;

NEURAL NETWORKS;

EID: 0025627940     PISSN: 08936080     EISSN: None     Source Type: Journal    
DOI: 10.1016/0893-6080(90)90005-6     Document Type: Article

References (27)

1. Approximation by superpositions of a sigmodial function
2. Fourier approximations and embeddings in Sobolev space
3. An elastivity can be estimated consistently without a priori knowledge of functional form
4. On the approximate realization of continuous mappings by neural networks
5. Identification and consistency in semi-non-parametric regression
6. On learning the derivatives of an unknown mapping with multilayer feedforward networks
7. A general explanation and interrogation system for neural networks. Poster Presentation
8. Theory of the back-propagation neural network
9. Multilayer feedforward networks are universal approximators
10. Capabilities of three-layered perceptrons
11. Generic constraints on underspecified target trajectories
12. Nonlinear signal processing using neural networks: Prediction and system modeling
13. Approximations of differentiable functions
14. Universal approximation using feedforward networks with non-sigmoid hidden layer activation functions
15. Non-parametric estimation of econometric functionals
16. Flexible production function estimation by non-parametric kernel estimators