Spectral Convergence of Random Feature Method in Multiple Dimensions

DOI10.48550/arXiv.2609.03401MaRDI QIDQ7285188arXivFDO

Authors Pingbing Ming, Hao Yu

Publication date 3 September 2026

Published in arXiv

Tldr We analyze convergence of random feature methods in multiple dimensions for several classes of targets, establish abstract error estimates for second-order elliptic problems, and show singular value decay with corresponding condition-number lower bounds for RFMtxs.

Full work available at URL https://doi.org/10.48550/arXiv.2609.03401

arXiv classification

cs.AIcs.LGmath.NAmath.ST

Cites work

• A Structure-Adaptive Random Feature Method for High-Dimensional Elliptic PDEs
• Adaptive random Fourier features with Metropolis sampling
• An introduction to the theory of reproducing kernel Hilbert spaces
• An Operator Pseudo-Inversion Lemma
• Approximation bounds for random neural networks and reservoir systems
• Approximation Theory and Applications of Randomized Neural Networks for Solving High-Dimensional PDEs
• Approximation variationnelle des problèmes aux limites
• Asymptotic Behavior of the Eigenvalues of Certain Integral Equations
• Asymptotic behavior of the eigenvalues of certain integral equations. II
• Bridging traditional and machine learning-based algorithms for solving PDEs: the random feature method
• Certain inequalities involving prolate spheroidal wave functions and associated quantities
• Certain upper bounds on the eigenvalues associated with prolate spheroidal wave functions
• Concentration of Random Feature Matrices in High-Dimensions
• Conditioning of random Fourier feature matrices: double descent and generalization error
• Deep ReLU Networks Overcome the Curse of Dimensionality for Generalized Bandlimited Functions
• Elliptic problems in nonsmooth domains
• Exponential convergence of the deep neural network approximation for analytic functions
• Extreme learning machine: Theory and applications
• Finite element approximation of eigenvalue problems
• Full error analysis of the random deep splitting method for nonlinear parabolic PDEs and PIDEs with infinite activity
• Global pseudo-differential calculus on Euclidean spaces
• High-precision randomized preconditioned iterative methods for the random feature method
• How exponentially ill-conditioned are contiguous submatrices of the Fourier matrix?
• Local Extreme Learning Machines and Domain Decomposition for Solving Linear and Nonlinear Partial Differential Equations
• Local Feature Filtering for Scalable and Well-Conditioned Domain-Decomposed Random Feature Methods
• Local Randomized Neural Networks with Discontinuous Galerkin Methods for Partial Differential Equations
• Mercer's theorem on general domains: on the interaction between measures, kernels, and RKHSs
• Notes on Exact Boundary Values in Residual Minimisation
• On Restrictions and Extensions of the Besov and Triebel-Lizorkin Spaces with Respect to Lipschitz Domains
• On some extensions of Bernstein's inequality for self-adjoint operators
• On the approximation of functions by tanh neural networks
• On the equivalence between kernel quadrature rules and random feature expansions
• Operator learning using random features: a tool for scientific computing
• Optimal Sobolev Approximation by Deterministic and Random Shallow Sigmoidal Networks
• Optimization of random feature method in the high-precision regime
• Overlapping Schwarz Preconditioners for Randomized Neural Networks with Domain Decomposition
• Polynomial approximation in Sobolev spaces on the unit sphere and the unit ball
• Random feature neural networks learn Black-Scholes type PDEs without curse of dimensionality
• Random projection neural networks of best approximation: convergence theory and practical applications
• Scattered Data Approximation
• Simultaneous approximation in Lebesgue and Sobolev norms via eigenspaces
• Simultaneous approximation of a function and its derivatives
• Spectral connvergece of random feature method in one dimension
• SSBE-PINN: A Sobolev Boundary Scheme Boosting Stability and Accuracy in Elliptic/Parabolic PDE Learning
• Sui problemi ai limiti i sistemi lineari ellittici
• The Gamma function
• The h, p and h-p versions of the finite element method in 1 dimension. II. The error analysis of the h- and h-p versions
• The Random Feature Method for Time-Dependent Problems
• The Random Feature Model for Input-Output Maps between Banach Spaces
• Towards a Mathematical Understanding of Neural Network-Based Machine Learning: what we know and what we don't
• Trace Regularity PINNs: Enforcing $\mathrm{H}^{\frac{1}{2}}(\partial Ω)$ for Boundary Data

Cited in

(1)

• Fully Deterministic Shallow Neural Networks with Localized Activations. Part I: Construction and Optimal Sobolev Approximation

This page was built for publication: Spectral Convergence of Random Feature Method in Multiple Dimensions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7285188)