Refine Now, Query Fast: A Decoupled Refinement Paradigm for Implicit Neural Fields

TL;DR

This paper introduces the Decoupled Representation Refinement (DRR) paradigm for implicit neural fields, achieving high fidelity and faster inference by decoupling complex neural networks from the inference process, validated by state-of-the-art results.

Contribution

The paper proposes a novel DRR architectural paradigm that separates high-capacity neural refinement from fast inference, enabling efficient and accurate implicit neural representations.

Findings

Achieves up to 27× faster inference than high-fidelity baselines.

Demonstrates state-of-the-art fidelity on ensemble simulation datasets.

Introduces the DRR-Net and Variational Pairs data augmentation strategy.

Abstract

Implicit Neural Representations (INRs) have emerged as promising surrogates for large 3D scientific simulations due to their ability to continuously model spatial and conditional fields, yet they face a critical fidelity-speed dilemma: deep MLPs suffer from high inference cost, while efficient embedding-based models lack sufficient expressiveness. To resolve this, we propose the Decoupled Representation Refinement (DRR) architectural paradigm. DRR leverages a deep refiner network, alongside non-parametric transformations, in a one-time offline process to encode rich representations into a compact and efficient embedding structure. This approach decouples slow neural networks with high representational capacity from the fast inference path. We introduce DRR-Net, a simple network that validates this paradigm, and a novel data augmentation strategy, Variational Pairs (VP) for improving INRs under complex tasks like high-dimensional surrogate modeling. Experiments on several ensemble simulation datasets demonstrate that our approach achieves state-of-the-art fidelity, while being up to 27$\times$ faster at inference than high-fidelity baselines and remaining competitive with the fastest models. The DRR paradigm offers an effective strategy for building powerful and practical neural field surrogates and INRs in broader applications, with a minimal compromise between speed and quality.