AI Paper Atlas

Research map / Graph learning / GNN architectures

Module-Aware Optimization for Graph Neural Networks

Guy Hadad, Haggai Roitman, Moshe Eliasof, Bracha Shapira

NeurIPS 2026 · Graph learning › GNN architectures · approach: Physics-inspired GNNs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a module-aware optimizer for GNNs that treats channel-mixing weights and graph-propagation components differently, rather than applying one update geometry to all parameters.

Problem

Optimizers such as Adam/AdamW (coordinate-wise) and Muon (orthogonalized matrix updates) impose fixed geometries on parameter updates, but in GNNs the effective geometry of weight matrices is also shaped by graph propagation, which these optimizers ignore.

Approach

Makes the optimizer aware of GNN module structure: updates are chosen per module, accounting for the graph-propagation operator as well as learned channel-mixing weights, instead of applying one geometry uniformly.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add graph-aware orthogonalization and step normalization to MuonMuon · ogbn-arxiv, ogbn-mag, ogbn-products (GCN)validation-loss time-to-target: 1.32x-2.42x speedup over AdamWImproved
add graph-aware orthogonalization alone to Muon (ablation)Muon · ogbn-arxivspeedup vs Muon: 1.00x -> 5.39xImproved
add graph-aware orthogonalization alone to Muon (ablation)Muon · ogbn-mag, ogbn-productsspeedup vs Muon: 1.07x and 1.03xNo clear effect
add graph-aware step normalization alone to Muon (ablation)Muon · ogbn-magspeedup vs Muon: 1.00x -> 2.28xImproved
combine graph-aware orthogonalization with step normalization (ablation)GraphMuon · ogbn-magspeedup vs Muon: 2.28x -> 3.60xImproved
use GraphMuon instead of AdamWGAT · ogbn-arxivtest accuracy: .6928 -> .6926No clear effect
use GraphMuon instead of AdamWGraphGPS · ogbn-magtest accuracy: .2939 -> .3296Improved
use GraphMuon instead of AdamWGINE · ZINCMAE: .2303 -> .1897Improved
use GraphMuon instead of AdamWGINE · ogbg-moltox21wall-clock speedup: 1.20x speedup, ROC-AUC .6633 -> .6644Mixed
use Muon instead of AdamWMuon · ogbn-products, ogbg-molbacetime-to-target: did not reach targetHurt

Experimental setup

Baselines: Adam AdamW Muon

Builds on: Muon Adam AdamW

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