English

F-INR: Functional Tensor Decomposition for Implicit Neural Representations

Machine Learning 2025-11-27 v2

Abstract

Implicit Neural Representations (INRs) model signals as continuous, differentiable functions. However, monolithic INRs scale poorly with data dimensionality, leading to excessive training costs. We propose F-INR, a framework that addresses this limitation by factorizing a high-dimensional INR into a set of compact, axis-specific sub-networks based on functional tensor decomposition. These sub-networks learn low-dimensional functional components that are then combined via tensor operations. This factorization reduces computational complexity while additionally improving representational capacity. F-INR is both architecture- and decomposition-agnostic. It integrates with various existing INR backbones (e.g., SIREN, WIRE, FINER, Factor Fields) and tensor formats (e.g., CP, TT, Tucker), offering fine-grained control over the speed-accuracy trade-off via the tensor rank and mode. Our experiments show F-INR accelerates training by up to 20×20\times and improves fidelity by over \num{6.0} dB PSNR compared to state-of-the-art INRs. We validate these gains on diverse tasks, including image representation, 3D geometry reconstruction, and neural radiance fields. We further show F-INR's applicability to scientific computing by modeling complex physics simulations. Thus, F-INR provides a scalable, flexible, and efficient framework for high-dimensional signal modeling. Project page: https://f-inr.github.io

Keywords

Cite

@article{arxiv.2503.21507,
  title  = {F-INR: Functional Tensor Decomposition for Implicit Neural Representations},
  author = {Sai Karthikeya Vemuri and Tim Büchner and Joachim Denzler},
  journal= {arXiv preprint arXiv:2503.21507},
  year   = {2025}
}

Comments

Accepted at WACV 2026. Website: https://f-inr.github.io Supplementary Material can be found there. 12 pages, 6 figures, 5 tables

R2 v1 2026-06-28T22:36:43.151Z