Neural Network Approximation: A View from Polytope Decomposition

• URL: http://arxiv.org/abs/2601.18264v1
• Date: Mon, 26 Jan 2026 08:39:11 GMT
• Title: Neural Network Approximation: A View from Polytope Decomposition
• Authors: ZeYu Li, ShiJun Zhang, TieYong Zeng, FengLei Fan,
• Abstract summary: We develop an explicit kernel method to derive an universal approximation of continuous functions.<n>A ReLU network is constructed to approximate the kernel in each subdomain separately.<n>We extend our approach to analytic functions to reach a higher approximation rate.
• Score: 36.24385738607139
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Universal approximation theory offers a foundational framework to verify neural network expressiveness, enabling principled utilization in real-world applications. However, most existing theoretical constructions are established by uniformly dividing the input space into tiny hypercubes without considering the local regularity of the target function. In this work, we investigate the universal approximation capabilities of ReLU networks from a view of polytope decomposition, which offers a more realistic and task-oriented approach compared to current methods. To achieve this, we develop an explicit kernel polynomial method to derive an universal approximation of continuous functions, which is characterized not only by the refined Totik-Ditzian-type modulus of continuity, but also by polytopical domain decomposition. Then, a ReLU network is constructed to approximate the kernel polynomial in each subdomain separately. Furthermore, we find that polytope decomposition makes our approximation more efficient and flexible than existing methods in many cases, especially near singular points of the objective function. Lastly, we extend our approach to analytic functions to reach a higher approximation rate.

Related papers

• Universal Approximation Theorem for Input-Connected Multilayer Perceptrons [0.0]
  We introduce the Input-Connected Multilayer Perceptron (IC-MLP), a feedforward neural network architecture in which each hidden neuron receives, in addition to the outputs of the preceding layer, a direct affine connection from the raw input.<n>We prove a universal approximation theorem showing that deep IC-MLPs can approximate any continuous function on a closed interval of the real line if and only if the activation function is nonlinear.<n>We then extend the analysis to vector-valued inputs and establish a corresponding universal approximation theorem for continuous functions on compact subsets of $mathbbRn$.
  arXiv  Detail & Related papers  (2026-01-20T14:48:45Z)
• DeePoly: A High-Order Accuracy Scientific Machine Learning Framework for Function Approximation and Solving PDEs [5.483488375189695]
  This work introduces a novel framework that transforms the Dee solution to a two-stage approach.<n>The strategic combination leverages the strengths of both methods.<n>This approach also serves as the open-source project also serves as the paper.
  arXiv  Detail & Related papers  (2025-06-05T04:10:52Z)
• Universal approximation results for neural networks with non-polynomial activation function over non-compact domains [3.3379026542599934]
  We extend the universal approximation property of single-hidden-layer feedforward neural networks beyond compact domains.<n>By assuming that the activation function is non-polynomial, we establish universal approximation results within function spaces defined over non-compact subsets of a Euclidean space.
  arXiv  Detail & Related papers  (2024-10-18T09:53:20Z)
• Stable Nonconvex-Nonconcave Training via Linear Interpolation [51.668052890249726]
  This paper presents a theoretical analysis of linearahead as a principled method for stabilizing (large-scale) neural network training. We argue that instabilities in the optimization process are often caused by the nonmonotonicity of the loss landscape and show how linear can help by leveraging the theory of nonexpansive operators.
  arXiv  Detail & Related papers  (2023-10-20T12:45:12Z)
• Active Learning-based Domain Adaptive Localized Polynomial Chaos Expansion [0.0]
  The paper presents a novel methodology to build surrogate models of complicated functions by an active learning-based sequential decomposition of the input random space and construction of localized chaos expansions. The approach utilizes sequential decomposition of the input random space into smaller sub-domains approximated by low-order expansions.
  arXiv  Detail & Related papers  (2023-01-31T13:49:52Z)
• The Dynamics of Riemannian Robbins-Monro Algorithms [101.29301565229265]
  We propose a family of Riemannian algorithms generalizing and extending the seminal approximation framework of Robbins and Monro. Compared to their Euclidean counterparts, Riemannian algorithms are much less understood due to lack of a global linear structure on the manifold. We provide a general template of almost sure convergence results that mirrors and extends the existing theory for Euclidean Robbins-Monro schemes.
  arXiv  Detail & Related papers  (2022-06-14T12:30:11Z)
• Quantitative Rates and Fundamental Obstructions to Non-Euclidean Universal Approximation with Deep Narrow Feed-Forward Networks [3.8073142980733]
  We quantify the number of narrow layers required for "deep geometric feed-forward neural networks" We find that both the global and local universal approximation guarantees can only coincide when approximating null-homotopic functions.
  arXiv  Detail & Related papers  (2021-01-13T23:29:40Z)
• Universal Approximation Property of Neural Ordinary Differential Equations [19.861764482790544]
  We show that NODEs can form an $Lp$-universal approximator for continuous maps under certain conditions. We also show their stronger approximation property, namely the $sup$-universality for approximating a large class of diffeomorphisms.
  arXiv  Detail & Related papers  (2020-12-04T05:53:21Z)
• Multipole Graph Neural Operator for Parametric Partial Differential Equations [57.90284928158383]
  One of the main challenges in using deep learning-based methods for simulating physical systems is formulating physics-based data. We propose a novel multi-level graph neural network framework that captures interaction at all ranges with only linear complexity. Experiments confirm our multi-graph network learns discretization-invariant solution operators to PDEs and can be evaluated in linear time.
  arXiv  Detail & Related papers  (2020-06-16T21:56:22Z)
• Polynomial-Time Exact MAP Inference on Discrete Models with Global Dependencies [83.05591911173332]
  junction tree algorithm is the most general solution for exact MAP inference with run-time guarantees. We propose a new graph transformation technique via node cloning which ensures a run-time for solving our target problem independently of the form of a corresponding clique tree.
  arXiv  Detail & Related papers  (2019-12-27T13:30:29Z)
• Neural Proximal/Trust Region Policy Optimization Attains Globally Optimal Policy [119.12515258771302]
  We show that a variant of PPOO equipped with over-parametrization converges to globally optimal networks. The key to our analysis is the iterate of infinite gradient under a notion of one-dimensional monotonicity, where the gradient and are instant by networks.
  arXiv  Detail & Related papers  (2019-06-25T03:20:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.