AI Paper Atlas

Research map / Graph learning / GNN architectures

Rethinking Graph Transformers as Graph Signal Denoisers: The Role of Block-Diagonal Priors

Jiaming Zhuo, Ziyi Ma, Kun Fu, Di Jin, Chuan Wang, Zhen Wang, Xiaochun Cao, Huazhu Fu, Liang Yang

ICML 2026 · Graph learning › GNN architectures · approach: GNN generalization theory

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 BDFormer block-diagonal anchor propagation instead of SGFormerGraph Transformer · Chameleonaccuracy: 43.77 -> 47.90Improved
use BDFormer instead of SGFormerGraph Transformer · Squirrelaccuracy: 41.09 -> 44.20Improved
use BDFormer instead of M3DphormerGraph Transformer · MinesweeperROC-AUC: 97.83 -> 97.35Hurt
use BDFormer instead of SGFormerGraph Transformer · ogbn-proteinsROC-AUC: 79.53 -> 80.31Improved
use BDFormer instead of PolynormerGraph Transformer · ogbn-arxivaccuracy: 73.46 -> 73.66No clear effect
use BDFormer instead of PolynormerGraph Transformer · ogbn-productsaccuracy: 83.82 -> 83.98Improved
use BDFormer instead of SGFormer on long-range benchmarkGraph Transformer · Parisaccuracy: 52.00 -> 59.01Improved
use BDFormer instead of Polynormer on long-range benchmarkGraph Transformer · Shanghaiaccuracy: 71.42 -> 71.69Improved
prune inter-class edges with learned anchor affinity (topology refinement)BDFormer · Chameleon / CiteSeer edge ratiosinter/intra-class edge ratioImproved
increase number of anchors k and blocks g beyond k>=64, g>=10 (ablation)BDFormer · Hyperparameter sweepaccuracy: gains saturate at k >= 64 and g >= 10No clear effect
vary input graph size from 20K to 100K nodesBDFormer · ogbn-arxivtraining time and GPU memory: linear growthNo clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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