Papers › Dimensionality Reduction Meets Message Passing for Graph Node Embeddings

Dimensionality Reduction Meets Message Passing for Graph Node Embeddings

1 Feb 2022arXiv:2202.00408archive 2025-07-28

Krzysztof Sadowski, Michał Szarmach, Eddie Mattia

Graph Neural Networks (GNNs) have become a popular approach for various applications, ranging from social network analysis to modeling chemical properties of molecules. While GNNs often show remarkable performance on public datasets, they can struggle to learn long-range dependencies in the data due to over-smoothing and over-squashing tendencies. To alleviate this challenge, we propose PCAPass, a method which combines Principal Component Analysis (PCA) and message passing for generating node embeddings in an unsupervised manner and leverages gradient boosted decision trees for classification tasks. We show empirically that this approach provides competitive performance compared to popular GNNs on node classification benchmarks, while gathering information from longer distance neighborhoods. Our research demonstrates that applying dimensionality reduction with message passing and skip connections is a promising mechanism for aggregating long-range dependencies in graph structured data.

PaperPDFCode

Code

ksadowski13/PCAPass officialpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Dimensionality ReductionNode ClassificationNode Property Prediction

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Node Classification Reddit PCAPass + XGBoost Accuracy 96.26 ± 0.02% #10 of 16 Archive leaderboard report
Node Property Prediction ogbn-arxiv PCAPass + XGBoost Ext. data No #75 of 86 Archive leaderboard report
Node Property Prediction ogbn-arxiv PCAPass + XGBoost Number of params 0 #75 of 86 Archive leaderboard report
Node Property Prediction ogbn-arxiv PCAPass + XGBoost Test Accuracy 0.7187 ± 0.0003 #75 of 86 Archive leaderboard report
Node Property Prediction ogbn-arxiv PCAPass + XGBoost Validation Accuracy 0.7325 ± 0.0005 #75 of 86 Archive leaderboard report
Node Property Prediction ogbn-papers100M PCAPass + LightGBM Ext. data No #16 of 20 Archive leaderboard report
Node Property Prediction ogbn-papers100M PCAPass + LightGBM Number of params 0 #16 of 20 Archive leaderboard report
Node Property Prediction ogbn-papers100M PCAPass + LightGBM Test Accuracy 0.6591 ± 0.0003 #16 of 20 Archive leaderboard report
Node Property Prediction ogbn-papers100M PCAPass + LightGBM Validation Accuracy 0.6982 ± 0.0002 #16 of 20 Archive leaderboard report
Node Property Prediction ogbn-products PCAPass + XGBoost Ext. data No #39 of 64 Archive leaderboard report
Node Property Prediction ogbn-products PCAPass + XGBoost Number of params 0 #39 of 64 Archive leaderboard report
Node Property Prediction ogbn-products PCAPass + XGBoost Test Accuracy 0.8115 ± 0.0002 #39 of 64 Archive leaderboard report
Node Property Prediction ogbn-products PCAPass + XGBoost Validation Accuracy 0.9200 ± 0.0005 #39 of 64 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections