AI Paper Atlas

Research map / Graph learning / GNN architectures

Learning Physical Dynamics with Subequivariant Graph Neural Networks

Jiaqi Han, Wenbing Huang, Hengbo Ma, Jiachen Li, Joshua B. Tenenbaum, Chuang Gan

NeurIPS 2022 · Graph learning › GNN architectures · approach: Structural GNN variants · 10 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use subequivariance rather than full E(3)-equivariance (ablation)SGNN · Physion Dominoes, Contain, and Dropcontact prediction accuracy (%): 79.6 -> 89.1; 65.3 -> 78.1; 68.1 -> 74.1Improved
Add object-aware message passing (ablation)SGNN · Physion Dominoes, Contain, and Dropcontact prediction accuracy (%): 81.2 -> 89.1; 68.7 -> 78.1; 70.1 -> 74.1Improved
Add hierarchical modeling (ablation)SGNN · Physion Dominoes, Contain, and Dropcontact prediction accuracy (%): 83.5 -> 89.1; 72.5 -> 78.1; 71.3 -> 74.1Improved
Separate inter-object and inner-object edges (ablation)SGNN · Physion Dominoes, Contain, and Dropcontact prediction accuracy (%): 85.0 -> 89.1; 74.8 -> 78.1; 72.0 -> 74.1Improved
Train with rotation augmentation instead of relying on equivarianceGNS · Physion Dominoescontact prediction accuracy (%): 53.3 -> 74.7Improved
Rotate test data while training on original, directionally biased dataGNS · Physion Dominoescontact prediction accuracy (%): 78.6 -> 53.3Hurt
Replace standard EGNN with gravity-aware EGNN-SEGNN · Physion Dominoescontact prediction accuracy (%): 61.3 -> 72.0Improved
Use SGNN instead of the best baselinePhysical dynamics simulators · RigidFallrollout MSE (× 10⁻²): DPI 1.71 -> SGNN 0.72 (|Train|=200, t=20); DPI 5.37 -> SGNN 2.44 (|Train|=200, t=40)Improved
Use SGNN instead of GNSPhysical dynamics simulators · Physioncontact prediction accuracy (%): 78.6 -> 89.1 (Dominoes, GNS*); 70.7 -> 78.1 (Contain, GNS); 58.0 -> 73.3 (Link, GNS)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy