Generative Refinement Learning for Temporal Interpolation in Scientific Machine Learning

📄 arXiv: 2610.36751v1 📥 PDF

作者: Pu Ren, Michael W. Mahoney, N. Benjamin Erichson

分类: cs.LG, cs.AI, math-ph, stat.ML

发布日期: 2026-10-27

备注: Accepted by Transactions on Machine Learning Research

🔗 代码/项目: GITHUB


💡 一句话要点

提出iFLEX以解决科学机器学习中的时间插值问题

🎯 匹配领域: 支柱三:空间感知与语义 (Perception & Semantics)

关键词: 时间插值 科学机器学习 条件扩散模型 重建误差 流体动力学 数据插值 模型训练

📋 核心要点

  1. 现有方法在低时间分辨率数据插值中存在误差,无法有效重建中间科学场。
  2. 论文提出的iFLEX模型通过加权平均和条件扩散模型学习残差,改善了插值精度。
  3. iFLEX在Navier-Stokes湍流、上海雷达降水和海表温度等数据集上表现优异,重建误差显著降低。

📝 摘要(中文)

计算密集型模拟和现实世界观测通常提供的时间分辨率低于下游任务的需求。在固定网格点上平均观测值可能引入误差,单靠运动估计无法确定物理场值的变化。本文研究了学习修正简单数值插值是否能改善中间科学场的重建。我们提出的模型iFLEX通过对每个网格点的两个观测值进行加权平均来估计缺失的场,并在训练过程中利用条件扩散模型学习生成残差。实验结果表明,iFLEX在多个数据集上实现了比现有基线更低的重建误差。

🔬 方法详解

问题定义:本文旨在解决低时间分辨率数据插值中的重建误差问题。现有方法如固定网格点平均和运动估计无法准确捕捉物理场的变化,导致重建效果不佳。

核心思路:iFLEX模型通过对两个观测值进行加权平均来估计缺失的中间场,并利用条件扩散模型学习生成残差,从而提高重建精度。这样的设计使得模型能够更好地利用时间信息和物理参数。

技术框架:iFLEX的整体架构包括两个主要阶段:首先,计算加权平均以估计缺失场;其次,利用条件扩散模型生成残差。模型的输入包括两个观测值、目标时间、观测间隔和物理参数。

关键创新:iFLEX的创新之处在于引入了条件扩散模型来学习残差,这一方法与传统的插值方法有本质区别,能够更准确地重建中间场。

关键设计:模型的关键设计包括加权平均的计算方法、条件扩散模型的训练过程,以及损失函数的选择,确保模型在训练时能够有效学习到残差特征。通过固定扩散骨干和训练设置,模型在不同数据集上表现出色。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

在控制湍流比较中,iFLEX的学习残差方法将相对Frobenius范数误差从约13.7降低至10.5,显示出显著的性能提升。此外,iFLEX在多个数据集上均优于光流和扩散模型等基线方法,验证了其有效性。

🎯 应用场景

该研究的潜在应用领域包括气象预测、流体动力学模拟和其他需要高时间分辨率数据的科学计算。iFLEX能够有效提高中间场的重建精度,具有重要的实际价值和广泛的应用前景,尤其是在需要实时数据处理的场景中。

📄 摘要(原文)

Computationally expensive simulations and real-world observations often provide data at a lower temporal resolution than downstream tasks require. Averaging observations at fixed grid points can introduce errors, and motion estimates alone do not determine how the physical field values change. In this work, we study whether learning to correct a simple numerical interpolation improves the reconstruction of intermediate scientific fields. Our model, interpolative FLow EXpert (iFLEX), first estimates the missing field by taking a weighted average of the two observations at each grid point, giving more weight to the observation closer to the requested time. During training, subtracting this estimate from an observed intermediate field gives the residual that a conditional diffusion model learns to generate. Both observations condition the model, together with the target time, the interval between observations, and available physical parameters. To reconstruct an unobserved intermediate field, the generated residual is added to the weighted average. We evaluate iFLEX on Navier-Stokes Kraichnan turbulence, Shanghai radar precipitation, and sea surface temperature. Across these datasets, iFLEX achieves lower reconstruction error than the evaluated baselines, including optical-flow and diffusion models developed for video frame interpolation in computer vision. In a controlled turbulence comparison at a Reynolds number of 36,000, learning the residual reduces relative Frobenius norm error from about 13.7 for direct field generation to 10.5, with the diffusion backbone and training setup held fixed. Code is available at https://github.com/paulpuren/iflex.