AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Handling Missing Data with Graph Representation Learning

Jiaxuan You, Xiaobai Ma, Yi Ding, Mykel J Kochenderfer, Jure Leskovec

NeurIPS 2020 · Graph learning › Graph self-supervised learning · approach: Heterophily and imbalance on graphs · 93 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
GRAPE vs best baseline (KNN)feature imputation · 9 UCI datasets, 30% missingMAE: 20% lower MAEImproved
GRAPE vs best baseline (MICE)label prediction · 9 UCI datasets, 30% missingMAE: 10% lower MAEImproved
GRAPE vs baselines at 10% missingfeature imputation · UCI, 10% missingMAE: -8% (KNN better)Hurt
add edge dropout (ablation)GRAPE · 9 UCI imputationMAE: 0.171 -> 0.090 (concrete); -33% avgImproved
use Mean instead of Sum aggregation (ablation)GRAPE · 9 UCI imputationMAE: 0.134 -> 0.102 (power)Improved
use Mean instead of Max aggregation (ablation)GRAPE · 9 UCI imputationMAE: 0.088 vs 0.090 (concrete)No clear effect
end-to-end training instead of impute-then-predict (ablation)GRAPE · UCI label predictionMAE: 9.02 -> 4.72 (yacht); 0.004 -> 0.007 (naval)Mixed
GRAPE on unseen observations without retrainingfeature imputation · UCI unseen dataMAE: 21% lower MAEImproved
GRAPE vs specialized discrete matrix completion IGMCmatrix completion · FlixsterRMSE: 0.872 vs 0.899Hurt

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