AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Self-Supervised Node Representation Learning via Node-to-Neighbourhood Alignment

Wei Dong, Dawei Yan, Peng Wang

TPAMI 2024 · Graph learning › Graph self-supervised learning · approach: Contrastive learning with negatives and views · 9 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 topology-aware positive sampling (TAPS) instead of a random neighbour as the single positive (ablation)N2N (JL) · Coramicro-F1: 83.46 -> 85.46Improved
add TAPS instead of a random neighbour as the single positive (ablation)N2N (JL) · Pubmedmicro-F1: 86.20 -> 91.08Improved
add TAPS instead of a random neighbour as the single positive (ablation)N2N (URL) · Amazon Photomicro-F1: 86.07 -> 89.40Improved
sample more TAPS positives (1 to 5) (ablation)N2N (JL) · Coramicro-F1: 85.46 -> 87.10Improved
use full-neighbourhood N2N instead of one TAPS positive (ablation)N2N (JL) · Coauthor CSmicro-F1: 92.07 -> 93.06Improved
add whitening and decorrelation (GSD) constraint to MLP-MSE (ablation)NF-N2N (W) · Coramicro-F1: 36.30 -> 89.00Improved
add whitening and decorrelation (GSD) constraint to MLP-MSE (ablation)NF-N2N (W) · Com-Amazon (pure-structured)micro-F1: 44.25 -> 87.28Improved
replace InfoNCE negative-pair contrastive baseline with NF-N2N (W)NF-N2N · Cora Fullmicro-F1: 51.29 -> 61.76Improved
add auto- or cross-correlation regularisation to whitening (WA, WC) over W alone (ablation)NF-N2N · Coramicro-F1: 89.00 -> 88.80No clear effect
use NF-N2N (W) on feature-less large graphs vs MLP on adjacencyNF-N2N (W) · Com-Youtube (pure-structured)micro-F1: 38.25 -> 46.87Improved
use negative-free NF-N2N instead of contrastive baseline for memoryNF-N2N (W) · Coauthor PhysicsGPU memory (MB): 3,292 -> 2,552Improved
use MLP encoder with N2N instead of GNN encoders for training timeN2N-TAPS-1 (JL) · Coauthor Physicstime per epoch (ms): 11.31 -> 0.24Improved

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