Data to 7 Oct 2026

Graph networks as learnable physics engines for inference and control

arXiv 1806.01242 ↗First posted 4 Jun 2018cs.LG cs.AI stat.MLcited by 242

Not on the map. Why?

Abstract (arXiv)

Understanding and interacting with everyday physical scenes requires rich knowledge about the structure of the world, represented either implicitly in a value or policy function, or explicitly in a transition model. Here we introduce a new class of learnable models--based on graph networks--which implement an inductive bias for object- and relation-centric representations of complex, dynamical systems. Our results show that as a forward model, our approach supports accurate predictions from real and simulated data, and surprisingly strong and efficient generalization, across eight distinct physical systems which we varied parametrically and structurally. We also found that our inference model can perform system identification. Our models are also differentiable, and support online planning via gradient-based trajectory optimization, as well as offline policy optimization. Our framework offers new opportunities for harnessing and exploiting rich knowledge about the world, and takes a key step toward building machines with more human-like representations of the world.

Key facts. Cited by 242 papers in the map (data through 7 Oct 2026). First posted on arXiv 4 Jun 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2020: 1; 2021: 58; 2022: 67; 2023: 48; 2024: 25; 2025: 22; 2026: 21.

242 papers.

Sort:

By area:

2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,023
2023VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Modelscited by 733
2024ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulationcited by 305
2022GraphCast: Learning skillful medium-range global weather forecastingcited by 261
2024PhysGen: Rigid-Body Physics-Grounded Image-to-Video Generationcited by 169
2022SlotFormer: Unsupervised Visual Dynamics Simulation with Object-Centric Modelscited by 146
2021Physion: Evaluating Physical Prediction from Vision in Humans and Machinescited by 145
2022Learning Causally Invariant Representations for Out-of-Distribution Generalization on Graphscited by 144
2022Pre-training via Denoising for Molecular Property Predictioncited by 124
2022Revisiting Over-smoothing and Over-squashing Using Ollivier-Ricci Curvaturecited by 96
2021Learned Coarse Models for Efficient Turbulence Simulationcited by 90
2021Size-Invariant Graph Representations for Graph Classification Extrapolationscited by 82

38 more

2022Predicting Physics in Mesh-reduced Space with Temporal Attentioncited by 82
2022MetaMorph: Learning Universal Controllers with Transformerscited by 78
2022AirfRANS: High Fidelity Computational Fluid Dynamics Dataset for Approximating Reynolds-Averaged Navier-Stokes Solutionscited by 77
2021A Survey on Interpretable Reinforcement Learningcited by 70
2021Port-Hamiltonian Neural Networks for Learning Explicit Time-Dependent Dynamical Systemscited by 66
2022When Physics Meets Machine Learning: A Survey of Physics-Informed Machine Learningcited by 64
2022Learning Multi-Object Dynamics with Compositional Neural Radiance Fieldscited by 63
2022A Generalist Neural Algorithmic Learnercited by 58
2022GraphFramEx: Towards Systematic Evaluation of Explainability Methods for Graph Neural Networkscited by 57
2022Multimodal learning with graphscited by 54
2021Learning to schedule job-shop problems: Representation and policy learning using graph neural network and reinforcement learningcited by 52
2022Graph Inverse Reinforcement Learning from Diverse Videoscited by 45
2021Simple GNN Regularisation for 3D Molecular Property Prediction & Beyondcited by 44
2022DiffSkill: Skill Abstraction from Differentiable Physics for Deformable Object Manipulations with Toolscited by 44
2021Physics-Guided Deep Learning for Dynamical Systems: A Surveycited by 43
2022Efficient Learning of Mesh-Based Physical Simulation with BSMS-GNNcited by 43
2022Learning rigid dynamics with face interaction graph networkscited by 43
2021Simulating Continuum Mechanics with Multi-Scale Graph Neural Networkscited by 39
2021Efficient Differentiable Simulation of Articulated Bodiescited by 38
2022Learning Physical Dynamics with Subequivariant Graph Neural Networkscited by 38
2021Hamiltonian-based Neural ODE Networks on the SE(3) Manifold For Dynamics Learning and Controlcited by 37
2022Physical Design using Differentiable Learned Simulatorscited by 37
2023Learning Neural PDE Solvers with Parameter-Guided Channel Attentioncited by 37
2022SizeShiftReg: a Regularization Method for Improving Size-Generalization in Graph Neural Networkscited by 36
2022Distributed Graph Neural Network Training: A Surveycited by 32
2022Graph Lifelong Learning: A Surveycited by 30
2021Constructing Neural Network-Based Models for Simulating Dynamical Systemscited by 29
2023Joint Learning of Label and Environment Causal Independence for Graph Out-of-Distribution Generalizationcited by 29
2022MAgNet: Mesh Agnostic Neural PDE Solvercited by 28
2023Physics-Informed Machine Learning for Modeling and Control of Dynamical Systemscited by 28
2020Differentiable Physics Models for Real-world Offline Model-based Reinforcement Learningcited by 27
2021Constraint-based graph network simulatorcited by 27
2024Entity-Centric Reinforcement Learning for Object Manipulation from Pixelscited by 27
2023LagrangeBench: A Lagrangian Fluid Mechanics Benchmarking Suitecited by 26
2022AnyMorph: Learning Transferable Polices By Inferring Agent Morphologycited by 25
2021An Extensible Benchmark Suite for Learning to Simulate Physical Systemscited by 24
2023A Generalist Dynamics Model for Controlcited by 24
2024HAMLET: Graph Transformer Neural Operator for Partial Differential Equationscited by 24

50 most cited of 242 All 242 (JSON)