Sparsely connected rank-inspired neural network

DOI10.48550/arXiv.2609.30750MaRDI QIDQ7360702zbMATH OpenarXivFDO

Authors Yunqing Huang, Nianyu Yi, Peimeng Yin, Wu Su

Publication date 25 September 2026

Published in arXiv

Tldr We propose a Sparsely Connected Rank-Inspired Neural Network (SCRINN) for function approximation and the numerical solution of partial differential equations (PDEs), which constructs an approximately orthonormal neural basis while promoting progressive rank growth of its evaluation matrix and reducing trainable parameters by introducing structured sparse connectivity architecture and orthogonality- and normalization-guided loss function.

Full work available at URL https://doi.org/10.48550/arXiv.2609.30750

Mathematics Subject Classification ID

Algorithms for approximation of functions (65D15) Numerical methods for partial differential equations, boundary value problems (65N99) Artificial neural networks and deep learning (68T07)

arXiv classification

math.NA

Cites work

• A Subspace, Interior, and Conjugate Gradient Method for Large-Scale Bound-Constrained Minimization Problems
• An Interior Penalty Finite Element Method with Discontinuous Elements
• Approximation by superpositions of a sigmoidal function
• Deep collocation method: a framework for solving PDEs using neural networks with error control
• Discontinuous Galerkin Methods for Solving Elliptic and Parabolic Equations
• Extended Galerkin neural network approximation of singular variational problems with error control
• Extreme learning machine collocation for the numerical solution of elliptic PDEs with sharp gradients
• Extreme learning machine: Theory and applications
• Finite Volume Methods for Hyperbolic Problems
• Galerkin neural networks: a framework for approximating variational equations with error control
• Neural Galerkin schemes with active learning for high-dimensional evolution equations
• Numerical solution and bifurcation analysis of nonlinear partial differential equations with extreme learning machines
• On Computing the Hyperparameter of Extreme Learning Machines: Algorithm and Application to Computational PDEs, and Comparison with Classical and High-Order Finite Elements
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Rank-inspired neural network for solving partial differential equations
• The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
• The Finite Element Method for Elliptic Problems
• The Mathematical Theory of Finite Element Methods

This page was built for publication: Sparsely connected rank-inspired neural network

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7360702)