Networks can be considered as approximation schemes. Multilayer networks of the backpropagation type can approximate arbitrarily well continuous functions (Cybenko, 1989; Funahashi, 1989; Stinchcombe and White, 1989). We prove that networks derived from regularization theory and including Radial Basis Function (Poggio and Girosi, 1989), have a similar property. From the point of view of approximation theory, however, the property of approximating continous functions arbitrarily well is not sufficient for characterizing good approximation schemes. More critical is the property of best approximation. The main result of this paper is that multilayer networks, of the type used in backpropagation, are not best approximation. For regularization networks (in particular Radial Basis Function networks) we prove existence and uniqueness of best approximation.

Στην εργασία αυτή παρουσιάζονται τα βασικά στοιχεία της θεωρίας των συνεχιζόμενων κλασμάτων και στη συνέχεια αναπτύσσονται οι αλγόριθμοι που επιτρέπουν την εκτέλεση πράξεων μεταξύ συνεχιζόμενων κλασμάτων και ρητών αριθμώ,αλλά και μεταξύ συνεχιζόμενων κλασμάτων. / In this thesis we describe the basic theory of continued fractions and describe the algorithms that enable us to perform arithmetic operations with continued fractions and rational numbers,as well as with continued fractions.

Robust algorithms for the approximation of functions are studied and developed in this thesis. Novel results and algorithms on piecewise polynomial interpolation, rational interpolation and best polynomial and rational approximations are presented. Algorithms for the extension of Chebfun, a software system for the numerical computation with functions, are described. These algorithms allow the construction and manipulation of piecewise smooth functions numerically with machine precision. Breakpoints delimiting subintervals are introduced explicitly, implicitly or automatically, the latter method combining recursive subdivision and edge detection techniques. For interpolation by rational functions with free poles, a novel method is presented. When the interpolation nodes are roots of unity or Chebyshev points the algorithm is particularly simple and relies on discrete Fourier transform matrices, which results in a fast implementation using the Fast Fourier Transform. The method is generalised for arbitrary grids, which requires the construction of polynomials orthogonal on the set of interpolation nodes. The new algorithm has connections with other methods, particularly the work of Jacobi and Kronecker, Berrut and Mittelmann, and Egecioglu and Koc. Computed rational interpolants are compared with the behaviour expected from the theory of convergence of these approximants, and the difficulties due to truncated arithmetic are explained. The appearance of common factors in the numerator and denominator due to finite precision arithmetic is characterised by the behaviour of the singular values of the linear system associated with the rational interpolation problem. Finally, new Remez algorithms for the computation of best polynomial and rational approximations are presented. These algorithms rely on interpolation, for the computation of trial functions, and on Chebfun, for the location of trial references. For polynomials, the algorithm is particularly robust and efficient, and we report experiments with degrees in the thousands. For rational functions, we clarify the numerical issues that affect its application.

Neste trabalho, primeiramente, exploramos certos módulos de suavidade e K - funcionais definidos na esfera unitária m - dimensional e suas propriedades, dando prioridade a suas equivalências assintóticas e comparação com o erro de melhor aproximação. Uma das principais referências utilizadas foi (DAI; XU, 2010). Posteriormente, consideramos um módulo de suavidade e um K-funcional em espaços mais gerais, os espaços compactos 2-homogêneos, classe de espaços esta que contém a classe das esferas. A relação entre estes objetos e o raio de aproximação do operador translação (translação esférica, no contexto esférico) foi estudada. As principais referências foram (PLATONOV, 2009) e (PLATONOV, 1997). / In this work, we firstly explored certain moduli of smoothness and K - functionals defined on the m-dimensional unit sphere and their properties, mainly their asymptotic equivalence and relation to the best approximation error. The main reference is (DAI; XU, 2010). Later we consider a moduli of smoothness and a K-functional on a general setting, namely two-point homogeneous spaces, which has the unit spheres as one of its classes. Relations between those tools and the rate of approximation of the shiffting operator were studied. The main references here were (PLATONOV, 2009) and (PLATONOV, 1997).

Submitted by Carlos Alberto Do Nascimento (prof.math.edu@gmail.com) on 2018-06-04T20:15:40Z No. of bitstreams: 1 Dissertação.pdf: 2749900 bytes, checksum: d578113faba2ff3354a375ae6c4e7e1e (MD5) / Approved for entry into archive by Ana Paula Santulo Custódio de Medeiros null (asantulo@rc.unesp.br) on 2018-06-05T12:25:11Z (GMT) No. of bitstreams: 1 nascimento_ca_me_rcla.pdf: 2739624 bytes, checksum: c8831bbd095a228cff3121b4621d7091 (MD5) / Made available in DSpace on 2018-06-05T12:25:12Z (GMT). No. of bitstreams: 1 nascimento_ca_me_rcla.pdf: 2739624 bytes, checksum: c8831bbd095a228cff3121b4621d7091 (MD5) Previous issue date: 2018-05-03 / Neste trabalho, abordaremos os principais conceitos e propriedades sobre espaço de Banach e espaço de Hilbert com o objetivo de oferecer o conteúdo necessário para discutirmos algumas aplicações desses conceitos. Mostraremos a existência e unicidade de solução de Equações Diferenciais Ordinárias de Primeira Ordem, existência e unicidade de solução de certas Equações Integrais e existência e unicidade de melhor aproximação em espaços normados e de Hilbert. / In this work, we will discuss the main concepts and properties on Banach space and Hilbert space in order to offer the necessary content to discuss some applications of these concepts. We will show the existence and uniqueness of the solution of First Order Ordinary Differential Equations, existence and uniqueness of solution of certain Integral Equations and existence and uniqueness of better approximation in normed and Hilbert spaces.

Neste trabalho, primeiramente, exploramos certos módulos de suavidade e K - funcionais definidos na esfera unitária m - dimensional e suas propriedades, dando prioridade a suas equivalências assintóticas e comparação com o erro de melhor aproximação. Uma das principais referências utilizadas foi (DAI; XU, 2010). Posteriormente, consideramos um módulo de suavidade e um K-funcional em espaços mais gerais, os espaços compactos 2-homogêneos, classe de espaços esta que contém a classe das esferas. A relação entre estes objetos e o raio de aproximação do operador translação (translação esférica, no contexto esférico) foi estudada. As principais referências foram (PLATONOV, 2009) e (PLATONOV, 1997). / In this work, we firstly explored certain moduli of smoothness and K - functionals defined on the m-dimensional unit sphere and their properties, mainly their asymptotic equivalence and relation to the best approximation error. The main reference is (DAI; XU, 2010). Later we consider a moduli of smoothness and a K-functional on a general setting, namely two-point homogeneous spaces, which has the unit spheres as one of its classes. Relations between those tools and the rate of approximation of the shiffting operator were studied. The main references here were (PLATONOV, 2009) and (PLATONOV, 1997).