NN-approach to design of the optimal stochastic approximation algorithms

We develop a direct approach to the problem which leans upon neural network-based approximation to the optimal transformation function. This approach utilizes the following inherent properties of neural networks (NN): 1) the ability to produce an approximation to any continuous function with arbitra...

Full description

Saved in:
Bibliographic Details
Main Authors: Nazin, A.V, Shcherbakov, P.S
Format: Conference Proceeding
Language:English
Subjects:
Interpolation and function approximation (numerical analysis)
Mathematics computing
Neural computing techniques
Optimisation techniques
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We develop a direct approach to the problem which leans upon neural network-based approximation to the optimal transformation function. This approach utilizes the following inherent properties of neural networks (NN): 1) the ability to produce an approximation to any continuous function with arbitrary degree of accuracy, and 2) the ability to tune adaptively their behaviour.
DOI:10.1049/cp:19940173