AI Paper Atlas

Research map / Graph learning / GNN architectures

Connectivity-Guided Sparsification of 2-FWL GNNs: Preserving Full Expressivity with Improved Efficiency

Rongqin Chen, Fan Mo, Pak Lon Ip, Shenghui Zhang, Dan Wu, Ye Li, Leong Hou U

AAAI 2026 · Graph learning › GNN architectures · approach: Expressive equivariant message passing · 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
add Co-Sparsify connectivity-guided sparsification to PPGN+RRWPPPGN+RRWP · ZINC-subsettest MAE: 0.055 -> 0.050Improved
add Co-Sparsify to PPGN+RRWPPPGN+RRWP · ZINC-Fulltest MAE: 0.020 -> 0.018Improved
add Co-Sparsify to PPGN+RRWPPPGN+RRWP · QM9 U0MAE: 0.0111 -> 0.0149Hurt
add Co-Sparsify to PPGN+RRWPPPGN+RRWP · QM9 GMAE: 0.01278 -> 0.01767Hurt
add Co-Sparsify to PPGN+RRWPPPGN+RRWP · QM9 HMAE: 0.01266 -> 0.01637Hurt
add Co-Sparsify to PPGN+RRWPPPGN+RRWP · QM9 CvMAE: 0.08589 -> 0.07619Improved
add Co-Sparsify to PPGNPPGN · Substructure counting (8 targets)normalized MAE: 3-Cycles 0.0003 -> 0.0004; Tri.-Rec. 0.0144 -> 0.0049Mixed
add Co-Sparsify to PPGN (TUD classification vs UnionGNNs)PPGN · TUD FRANKENSTEINaccuracy (%): 68.02 -> 77.65Improved
add Co-Sparsify to PPGNPPGN · ZINC-subsetruntime per epoch (s): 9.3s -> 7.9sImproved
add Co-Sparsify to PPGNPPGN · QM9runtime per epoch (s): 97.1s -> 60.7sImproved
add Co-Sparsify to PPGNPPGN · QM9memory (GB): 6.4GB -> 4.2GBImproved
use standard CoSp-PPGN on long-range taskCoSp-PPGN · Peptides-structMAEHurt

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