mlsanityData to 14 Oct 2026

Architecture Shapes Transfer Specificity in Implicit Neural Representations

arXiv 2606.06827 ↗First posted 5 Jun 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Transfer in coordinate networks is often measured by warm-start gain, but whether that gain reflects source-specific structure or generic weight reuse is less clear. We study this question across three implicit neural representation (INR) families, SIREN, ReLU MLPs, and Fourier-feature MLPs, using controlled analytic tests, a 2D lid-driven-cavity Navier--Stokes benchmark, and 1D PDE reference-solution suites for heat, viscous Burgers, and focusing cubic NLS. The analytic tests use independent-seed random controls, while the PDE benchmarks use alternate same-family source controls and auxiliary ablations. Across settings, transfer magnitude and transfer specificity separate clearly. In a 10-seed controlled 1D geometric test, Fourier Features show the largest structured transfer (33.1×), followed by SIREN (23.0×) and ReLU (10.7×), but ReLU is far more selective: random-control transfer is 0.41× for ReLU versus 14.24× for SIREN. On a controlled two-parameter 1D family, the ranking changes: ReLU gives the clearest structured-versus-control separation at default settings, whereas Fourier Features improve only after bandwidth retuning. In Navier--Stokes and the broader 1D PDE suite, no single architecture dominates every equation, yet the same pattern remains: SIREN often reuses weights broadly, whereas ReLU and, in some equations, Fourier Features are more source-selective. Static diagnostics remain weak, and the heuristic scaling law Atransfer propto 1/Delta t2 is rejected in the implemented 1D audit. These results position transfer specificity as a useful diagnostic for coordinate networks and suggest that architecture selection in scientific machine learning should be evaluated under explicit control conditions, not by transfer magnitude alone.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 26 of its 43 references that are in the map. First posted on arXiv 5 Jun 2026.

Builds on

26 of 43 references.

References · 26

  1. 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,723
  2. 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,908
  3. 2020Implicit Neural Representations with Periodic Activation Functionscited by 2,238
  4. 2020Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domainscited by 2,200
  5. 2019DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operatorscited by 2,018
  6. 2020Sharpness-Aware Minimization for Efficiently Improving Generalizationcited by 1,417
  7. 2021Learning the solution operator of parametric partial differential equations with physics-informed DeepOnetscited by 540
  8. 2022Fourier Neural Operator with Learned Deformations for PDEs on General Geometriescited by 427
  9. 2021Factorized Fourier Neural Operatorscited by 249
  10. 2023FINER: Flexible spectral-bias tuning in Implicit NEural Representation by Variable-periodic Activation Functionscited by 78
  11. 2022Continuous PDE Dynamics Forecasting with Implicit Neural Representationscited by 73
  12. 2021Meta-Learning Sparse Implicit Neural Representationscited by 48
14 more
  1. 2024Where Do We Stand with Implicit Neural Representations? A Technical and Performance Surveycited by 42
  2. 2024Parameterized Physics-informed Neural Networks for Parameterized PDEscited by 38
  3. 2022Neural Implicit Flow: a mesh-agnostic dimensionality reduction paradigm of spatio-temporal datacited by 37
  4. 2023GPT-PINN: Generative Pre-Trained Physics-Informed Neural Networks toward non-intrusive Meta-learning of parametric PDEscited by 20
  5. 2025Transfer Learning in Physics-Informed Neural Networks: Full Fine-Tuning, Lightweight Fine-Tuning, and Low-Rank Adaptationcited by 15
  6. 2023Hyb-NeRF: A Multiresolution Hybrid Encoding for Neural Radiance Fieldscited by 6
  7. 2024A physics-informed transformer neural operator for learning generalized solutions of initial boundary value problemscited by 6
  8. 2023RANS-PINN based Simulation Surrogates for Predicting Turbulent Flowscited by 5
  9. 2024Physics-Aware Neural Implicit Solvers for multiscale, parametric PDEs with applications in heterogeneous mediacited by 5
  10. 2025A Physics-Informed Meta-Learning Framework for the Continuous Solution of Parametric PDEs on Arbitrary Geometriescited by 5
  11. 2024Physics-informed reduced order model with conditional neural fieldscited by 4
  12. 2026ArGEnT: Arbitrary Geometry-encoded Transformer for Operator Learningcited by 3
  13. 2025A New Perspective To Understanding Multi-resolution Hash Encoding For Neural Fieldscited by 2
  14. 2025Improving Accuracy and Efficiency of Implicit Neural Representations: Making SIREN a WINNERcited by 1

Built on by

Not cited yet.