Dependence of Computational Models on Input Dimension: Tractability of Approximation and Optimization Tasks

Authors: P.C. Kainen, Věra Kůrková, M. Sanguineti
Publication year: 2012
Publication type: Článek v odborném periodiku
Periodical: IEEE Transactions on Information Theory
Projects: Aproximation and learning of multivariable functions by neural networks
Complexity of perceptron and kernel networks
Learning of functional relationships from high-dimensional data
Bibliographic citation: KAINEN, P.C., KŮRKOVÁ, Věra, SANGUINETI, M. Dependence of Computational Models on Input Dimension: Tractability of Approximation and Optimization Tasks. IEEE Transactions on Information Theory. 2012, 58(2), 1203-1214. ISSN 0018-9448. E-ISSN 1557-9654
Dostupné z: doi:10.1109/TIT.2011.2169531
DOI: 10.1109/TIT.2011.2169531
ID: 368521

Abstract

The role of input dimension is studied in approximating, in various norms, target sets of d-variable functions using linear combinations of adjustable computational units. Results are applied to approximation and solution of optimization problems by neural networks with perceptron and Gaussian radial computational units.