-
Graph Neural PDE Solvers with Conservation and Similarity-Equivariance
Authors:
Masanobu Horie,
Naoto Mitsume
Abstract:
Utilizing machine learning to address partial differential equations (PDEs) presents significant challenges due to the diversity of spatial domains and their corresponding state configurations, which complicates the task of encompassing all potential scenarios through data-driven methodologies alone. Moreover, there are legitimate concerns regarding the generalization and reliability of such appro…
▽ More
Utilizing machine learning to address partial differential equations (PDEs) presents significant challenges due to the diversity of spatial domains and their corresponding state configurations, which complicates the task of encompassing all potential scenarios through data-driven methodologies alone. Moreover, there are legitimate concerns regarding the generalization and reliability of such approaches, as they often overlook inherent physical constraints. In response to these challenges, this study introduces a novel machine-learning architecture that is highly generalizable and adheres to conservation laws and physical symmetries, thereby ensuring greater reliability. The foundation of this architecture is graph neural networks (GNNs), which are adept at accommodating a variety of shapes and forms. Additionally, we explore the parallels between GNNs and traditional numerical solvers, facilitating a seamless integration of conservative principles and symmetries into machine learning models. Our findings from experiments demonstrate that the model's inclusion of physical laws significantly enhances its generalizability, i.e., no significant accuracy degradation for unseen spatial domains while other models degrade. The code is available at https://github.com/yellowshippo/fluxgnn-icml2024.
△ Less
Submitted 25 May, 2024;
originally announced May 2024.
-
2D Slice-driven Physics-based 3D Motion Estimation Framework for Pancreatic Radiotherapy
Authors:
Yuki Hara,
Noriyuki Kadoya,
Naoto Mitsume,
Naoto Ienaga,
Rei Umezawa,
Keiichi **gu,
Yoshihiro Kuroda
Abstract:
Pancreatic diseases are difficult to treat with high doses of radiation, as they often present both periodic and aperiodic deformations. Nevertheless, we expect that these difficulties can be overcome, and treatment results may be improved with the practical use of a device that can capture 2D slices of organs during irradiation. However, since only a few 2D slices can be taken, the 3D motion need…
▽ More
Pancreatic diseases are difficult to treat with high doses of radiation, as they often present both periodic and aperiodic deformations. Nevertheless, we expect that these difficulties can be overcome, and treatment results may be improved with the practical use of a device that can capture 2D slices of organs during irradiation. However, since only a few 2D slices can be taken, the 3D motion needs to be estimated from partially observed information. In this study, we propose a physics-based framework for estimating the 3D motion of organs, regardless of periodicity, from motion information obtained by 2D slices in one or more directions and a regression model that estimates the accuracy of the proposed framework to select the optimal slice. Using information obtained by slice-to-slice registration and setting the surrounding organs as boundaries, the framework drives the physical models for estimating 3D motion. The R2 score of the proposed regression model was greater than 0.9, and the RMSE was 0.357 mm. The mean errors were 5.11 $\pm$ 1.09 mm using an axial slice and 2.13 $\pm$ 0.598 mm using concurrent axial, sagittal, and coronal slices. Our results suggest that the proposed framework is comparable to volume-to-volume registration, and is feasible.
△ Less
Submitted 12 October, 2023;
originally announced October 2023.
-
Physics-Embedded Neural Networks: Graph Neural PDE Solvers with Mixed Boundary Conditions
Authors:
Masanobu Horie,
Naoto Mitsume
Abstract:
Graph neural network (GNN) is a promising approach to learning and predicting physical phenomena described in boundary value problems, such as partial differential equations (PDEs) with boundary conditions. However, existing models inadequately treat boundary conditions essential for the reliable prediction of such problems. In addition, because of the locally connected nature of GNNs, it is diffi…
▽ More
Graph neural network (GNN) is a promising approach to learning and predicting physical phenomena described in boundary value problems, such as partial differential equations (PDEs) with boundary conditions. However, existing models inadequately treat boundary conditions essential for the reliable prediction of such problems. In addition, because of the locally connected nature of GNNs, it is difficult to accurately predict the state after a long time, where interaction between vertices tends to be global. We present our approach termed physics-embedded neural networks that considers boundary conditions and predicts the state after a long time using an implicit method. It is built based on an E(n)-equivariant GNN, resulting in high generalization performance on various shapes. We demonstrate that our model learns flow phenomena in complex shapes and outperforms a well-optimized classical solver and a state-of-the-art machine learning model in speed-accuracy trade-off. Therefore, our model can be a useful standard for realizing reliable, fast, and accurate GNN-based PDE solvers. The code is available at https://github.com/yellowshippo/penn-neurips2022.
△ Less
Submitted 23 March, 2023; v1 submitted 24 May, 2022;
originally announced May 2022.
-
Isometric Transformation Invariant and Equivariant Graph Convolutional Networks
Authors:
Masanobu Horie,
Naoki Morita,
Toshiaki Hishinuma,
Yu Ihara,
Naoto Mitsume
Abstract:
Graphs are one of the most important data structures for representing pairwise relations between objects. Specifically, a graph embedded in a Euclidean space is essential to solving real problems, such as physical simulations. A crucial requirement for applying graphs in Euclidean spaces to physical simulations is learning and inferring the isometric transformation invariant and equivariant featur…
▽ More
Graphs are one of the most important data structures for representing pairwise relations between objects. Specifically, a graph embedded in a Euclidean space is essential to solving real problems, such as physical simulations. A crucial requirement for applying graphs in Euclidean spaces to physical simulations is learning and inferring the isometric transformation invariant and equivariant features in a computationally efficient manner. In this paper, we propose a set of transformation invariant and equivariant models based on graph convolutional networks, called IsoGCNs. We demonstrate that the proposed model has a competitive performance compared to state-of-the-art methods on tasks related to geometrical and physical simulation data. Moreover, the proposed model can scale up to graphs with 1M vertices and conduct an inference faster than a conventional finite element analysis, which the existing equivariant models cannot achieve.
△ Less
Submitted 10 March, 2021; v1 submitted 13 May, 2020;
originally announced May 2020.
