AI Paper Atlas

Research map / Graph learning / GNN architectures

Reversible and irreversible bracket-based dynamics for deep graph neural networks

Anthony Gruber, Kookjin Lee, Nathaniel Trask

NeurIPS 2023 · Graph learning › GNN architectures · approach: Physics-inspired GNNs · 1 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
replace black-box NODE with Hamiltonian bracketNODE · Damped double pendulumtotal MAE: 0.0269 -> 0.00369Improved
use gradient (purely dissipative) bracket instead of Hamiltonianbracket network · Damped double pendulumtotal MAE: 0.00369 -> 0.0208Hurt
add feature autoencoder to NODENODE · Damped double pendulumtotal MAE: 0.0269 -> 0.0602Hurt
use Hamiltonian bracket instead of NODE+AENODE+AE · MuJoCo HalfCheetahrelative error: 0.106 -> 0.0566Improved
use gradient bracket instead of NODE+AENODE+AE · MuJoCo HalfCheetahrelative error: 0.106 -> 0.105No clear effect
use metriplectic bracket instead of NODE+AENODE+AE · MuJoCo Swimmerrelative error: 0.0297 -> 0.0179Improved
use partially reversible double bracket instead of gradient bracketgradient bracket · Computer (random split)test accuracy: 78.1 -> 84.8Improved
use double bracket instead of gradient bracketgradient bracket · Photo (random split)test accuracy: 88.2 -> 92.4Improved
use Hamiltonian (fully reversible) bracket instead of gradient bracketgradient bracket · Cora (Planetoid)test accuracy: 79.9 -> 77.2Hurt
use metriplectic bracket instead of double bracketdouble bracket · Cora (Planetoid)test accuracy: 82.6 -> 57.4Hurt
double bracket vs GDEGDE · Cora (Planetoid)test accuracy: 83.8 -> 82.6No clear effect
double bracket vs GRAND-nlGRAND-nl · CiteSeer (Planetoid)test accuracy: 70.8 -> 74.2Improved

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