arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2401.11037cs.LGcs.NAmath.NAq-bio.QM
收藏

用于建模3D动力学的等变图神经算子

Equivariant Graph Neural Operator for Modeling 3D Dynamics

Stanford University(斯坦福大学);NVIDIA(英伟达);Argonne National Laboratory(阿贡国家实验室);California Institute of Technology(加州理工学院)

Minkai Xu, Jiaqi Han, Aaron Lou, Jean Kossaifi, Arvind Ramanathan, Kamyar Azizzadenesheli, Jure Leskovec, Stefano Ermon, Anima Anandkumar

AI总结:

针对现有图神经网络在3D动力学建模中无法捕捉时间相关性的问题,本文提出等变图神经算子(EGNO),通过在傅里叶空间参数化的等变时间卷积显式学习时间演化轨迹,首次在保持SE(3)等变性的同时实现解动力学函数建模,并在多领域实验中显著优于现有方法。

摘要:

建模关系系统的复杂三维(3D)动力学是自然科学中的一个重要问题,应用范围从分子模拟到粒子力学。机器学习方法通过学习图神经网络来建模空间交互,取得了良好的成功。然而,这些方法并不能忠实地捕捉时间相关性,因为它们仅对下一步预测进行建模。在本研究中,我们提出了Equivariant Graph Neural Operator (EGNO),这是一种新颖且有原则的方法,可直接将动力学建模为轨迹,而不仅仅是下一步预测。与现有方法不同,EGNO显式地学习3D动力学的时间演化,我们将动力学表述为关于时间的函数,并学习神经算子来近似它。为了在捕捉时间相关性的同时保持固有的SE(3)-等变性,我们开发了在傅里叶空间中参数化的等变时间卷积,并通过在等变网络上堆叠傅里叶层来构建EGNO。EGNO是第一个能够在随时间建模解动力学函数的同时保留3D等变性的算子学习框架。在多个领域的综合实验(包括粒子模拟、人体运动捕捉和分子动力学)表明,得益于等变时间建模,EGNO的性能显著优于现有方法。我们的代码可在https://github.com/MinkaiXu/egno获取。

英文摘要:

Modeling the complex three-dimensional (3D) dynamics of relational systems is an important problem in the natural sciences, with applications ranging from molecular simulations to particle mechanics. Machine learning methods have achieved good success by learning graph neural networks to model spatial interactions. However, these approaches do not faithfully capture temporal correlations since they only model next-step predictions. In this work, we propose Equivariant Graph Neural Operator (EGNO), a novel and principled method that directly models dynamics as trajectories instead of just next-step prediction. Different from existing methods, EGNO explicitly learns the temporal evolution of 3D dynamics where we formulate the dynamics as a function over time and learn neural operators to approximate it. To capture the temporal correlations while keeping the intrinsic SE(3)-equivariance, we develop equivariant temporal convolutions parameterized in the Fourier space and build EGNO by stacking the Fourier layers over equivariant networks. EGNO is the first operator learning framework that is capable of modeling solution dynamics functions over time while retaining 3D equivariance. Comprehensive experiments in multiple domains, including particle simulations, human motion capture, and molecular dynamics, demonstrate the significantly superior performance of EGNO against existing methods, thanks to the equivariant temporal modeling. Our code is available at https://github.com/MinkaiXu/egno.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑