Fast Amortized Fitting of Scientific Signals Across Time and Ensembles via Transferable Neural Fields

arXiv:2604.19979v1 Announce Type: cross Abstract: Neural fields, also known as implicit neural representations (INRs), offer a powerful framework for modeling continuous geometry, but their effectiveness in high-dimensional scientific settings is limited by slow convergence and scaling challenges. In this study, we extend INR models to handle spatiotemporal and multivariate signals and show how INR features can be transferred across scientific signals to enable efficient and scalable representation across time and ensemble runs in an amortized fashion. Across controlled transformation regimes (e.g., geometric transformations and localized perturbations of synthetic fields) and high-fidelity scientific domains-including turbulent flows, fluid-material impact dynamics, and astrophysical systems-we show that transferable features improve not only signal fidelity but also the accuracy of derived geometric and physical quantities, including density gradients and vorticity. In particular, transferable features reduce iterations to reach target reconstruction quality by up to an order of magnitude, increase early-stage reconstruction quality by multiple dB (with gains exceeding 10 dB in some cases), and consistently improve gradient-based physical accuracy.

Related

• NeuVolEx: Implicit Neural Features for Volume Exploration
• PCM-NeRF: Probabilistic Camera Modeling for Neural Radiance Fields under Pose Uncertainty
• Location Is All You Need: Continuous Spatiotemporal Neural Representations of Earth Observation Data
• Amortized Inverse Kinematics via Graph Attention for Real-Time Human Avatar Animation

Source: arXiv cs.CV | 2026-04-23