Function-space optimality of neural architectures with multivariate nonlinearities
From MaRDI portal
The authors introduce a family of function spaces defined with the \(k\)-plane transform, \(k\)-plane pseudodifferential operators, and filtering operators involving Laplacians of low dimensions. They present a representer theorem asserting that a solution to a regularization scheme for learning in such a function space is characterized by a shallow neural network with multivariate nonlinearities. These network architectures have skip connections and are connected to orthogonal weight normalization and multi-index models.
Cites work
- 10.1162/1532443041827925
- \(\alpha\)-Molecules
- A new algorithm for estimating the effective dimension-reduction subspace
- A unified Fourier slice method to derive ridgelet transform for a variety of depth-2 neural networks
- A unifying representer theorem for inverse problems and machine learning
- Analytic Tomography
- Approximation by superposition of sigmoidal and radial basis functions
- Approximation of generalized ridge functions in high dimensions
- Banach space representer theorems for neural networks and ridge splines
- Bounds on rates of variable-basis and neural-network approximation
- Breaking the curse of dimensionality with convex neural networks
- Capturing ridge functions in high dimensions from point queries
- Characterization of the variation spaces corresponding to shallow neural networks
- Convex optimization in sums of Banach spaces
- Degree of approximation by neural and translation networks with a single hidden layer
- Distributional extension and invertibility of the k-plane transform and its dual
- From kernel methods to neural networks: a unifying variational formulation
- scientific article; zbMATH DE number 5957198 (Why is no real title available?)
- scientific article; zbMATH DE number 438987 (Why is no real title available?)
- scientific article; zbMATH DE number 45848 (Why is no real title available?)
- scientific article; zbMATH DE number 3533448 (Why is no real title available?)
- scientific article; zbMATH DE number 1324223 (Why is no real title available?)
- scientific article; zbMATH DE number 1022519 (Why is no real title available?)
- scientific article; zbMATH DE number 3232443 (Why is no real title available?)
- scientific article; zbMATH DE number 3272562 (Why is no real title available?)
- scientific article; zbMATH DE number 3295108 (Why is no real title available?)
- scientific article; zbMATH DE number 3383043 (Why is no real title available?)
- scientific article; zbMATH DE number 3187905 (Why is no real title available?)
- Inversion of \(k\)-plane transforms via continuous wavelet transforms
- Inversion ofk-Plane Transforms and Applications in Computer Tomog
- Learning functions varying along a central subspace
- Multikernel regression with sparsity constraint
- Near-Minimax Optimal Estimation With Shallow ReLU Neural Networks
- On representer theorems and convex regularization
- On reproducing kernel Banach spaces: generic definitions and unified framework of constructions
- On the Range of the Radon d-Plane Transform and Its Dual
- On the tractability of multivariate integration and approximation by neural networks
- Practical and mathematical aspects of the problem of reconstructing objects from radiographs
- Scattered Data Approximation
- Sharp Bounds on the Approximation Rates, Metric Entropy, and n-Widths of Shallow Neural Networks
- Sliced Inverse Regression for Dimension Reduction
- Some results on Tchebycheffian spline functions and stochastic processes
- Sparsity of solutions for variational inverse problems with finite-dimensional data
- Spline solutions to L\(^1\) extremal problems in one and several variables
- Splines are universal solutions of linear inverse problems with generalized TV regularization
- The X-ray transform
- Understanding neural networks with reproducing kernel Banach spaces
- Weighted variation spaces and approximation by shallow ReLU networks
- What Kinds of Functions Do Deep Neural Networks Learn? Insights from Variational Spline Theory
- ℓ1 Regularization in Infinite Dimensional Feature Spaces
Cited in
(2)
This page was built for publication: Function-space optimality of neural architectures with multivariate nonlinearities
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7026068)
