AI Paper Atlas

Research map / Graph learning / GNN architectures

Label Deconvolution for Node Representation Learning on Large-Scale Attributed Graphs Against Learning Bias

Zhihao Shi, Jie Wang, Fanghua Lu, Hanzhu Chen, Defu Lian, Zheng Wang, Jieping Ye, Feng He Wu

TPAMI 2024 · Graph learning › GNN architectures · approach: Graph structure learning · 12 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 inverse labels (label deconvolution) instead of pseudo labels on true labels only (ablation)fine-tuned node encoder · ogbn-productaccuracy: 83.83 -> 86.45Improved
add inverse labels instead of pseudo labels (ablation)fine-tuned node encoder · ogbn-arxivaccuracy: 75.35 -> 77.26Improved
add inverse labels instead of pseudo labels (ablation)fine-tuned node encoder · ogbn-proteinROC-AUC: 89.15 -> 89.42Improved
add pseudo labels on evaluation nodes to true-label fine-tuning (ablation)X_FNE · ogbn-productaccuracy: 82.88 -> 83.83Improved
add pseudo labels on evaluation nodes to true-label fine-tuning (ablation)X_FNE · ogbn-arxivaccuracy: 75.52 -> 75.35No clear effect
replace GLEM with LD (single fine-tuning)separate training · ogbn-productaccuracy: 85.09 -> 86.45Improved
replace GLEM with LDseparate training · ogbn-producttraining time: 75.67h -> 24.78hImproved
use LD instead of fine-tuning with true labelsX_FNE · ogbn-productaccuracy: 82.88 -> 86.45Improved
use LD instead of fine-tuning with true labelsX_FNE · ogbn-producttraining time: 16.88h -> 24.78hHurt
use LD in inductive setting vs pre-trained embeddingsX_PNE · ogbn-arxiv inductiveaccuracy: 75.23 -> 76.83Improved
use LD in inductive setting vs GLEMX_GLEM · ogbn-protein inductiveROC-AUC: 88.72 -> 89.12Improved
shrink sampled subgraph size for joint trainingGCN+GAS · ogbn-arxivaccuracy: 74.28% -> 67.46%Hurt

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