Information-based nonlinear approximation: an average case setting
Given an element \(f\) of a Banach space \(X\) and a subset \(D\), called a dictionary, the authors study the so-called \(k\)-term approximations to \(f\), that is, approximations of the form \[ \sum_{j=1}^k a_jf_j \quad a_i \in {\mathbb R}, \quad f_j \in D. \] It is assumed that \(f\) is an element of a subset \(F \subset X\), equiped with a probability measure \(\mu\). The information about \(f\) is given by the values \(L_1f, \ldots , L_n f\) of some \(n\) linear functionals, and the error of approximation is estimated in the average (as opposed to the worst) case. It is shown that the problem can be essentially decomposed in two partial problems that can be solved independently. As an application, the authors consider piecewise polynomial approximation in \(C[0,1]\) on the class \(F_r\) of functions \(f \in C^r\) with \(\| f^{(r)}\| \leq 1\), with respect to the \(r\)-fold Wiener measure. In this case, to approximate \(f\) with error \(\varepsilon\) it is necessary and sufficient to know its values at \(O\left ( [\varepsilon^{-1}\ln^{1/2}(1/\varepsilon)]^{1/(r+1/2)} \right )\) equidistant points and use \(O\left (\varepsilon^{-1/(r+1/2)} \right )\) adaptively chosen breakpoints.
- Average case L_-approximation in the presence of Gaussian noise
- Complexity of neural network approximation with limited information: A worst case approach
- scientific article; zbMATH DE number 44104 (Why is no real title available?)
- scientific article; zbMATH DE number 1215245 (Why is no real title available?)
- scientific article; zbMATH DE number 1153816 (Why is no real title available?)
- Information of varying cardinality
- Probabilistic and average linear width in \(L_ \infty\)-norm with respect to \(r\)-fold Wiener measure
- The average errors for the Grünwald interpolation in the Wiener space
- Average case L_-approximation in the presence of Gaussian noise
- Adaptive wavelet methods for the stochastic Poisson equation
- Free-knot spline approximation of stochastic processes
- The optimal free knot spline approximation of stochastic differential equations with additive noise
- On linear versus nonlinear approximation in the average case setting
- Spline approximation of a random process with singularity
- Average errors for Kantorovitch operators on r-fold integrated Wiener space
- A Milstein-based free knot spline approximation for stochastic differential equations
- On the construction of stochastic fields with prescribed regularity by wavelet expansions
This page was built for publication: Information-based nonlinear approximation: an average case setting
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1772685)