Papers › FIT-GNN: Faster Inference Time for GNNs Using Coarsening
FIT-GNN: Faster Inference Time for GNNs Using Coarsening
Shubhajit Roy, Hrriday Ruparel, Kishan Ved, Anirban Dasgupta
Scalability of Graph Neural Networks (GNNs) remains a significant challenge, particularly when dealing with large-scale graphs. To tackle this, coarsening-based methods are used to reduce the graph into a smaller graph, resulting in faster computation. Nonetheless, prior research has not adequately addressed the computational costs during the inference phase. This paper presents a novel approach to improve the scalability of GNNs by reducing computational burden during both training and inference phases. We demonstrate two different methods (Extra-Nodes and Cluster-Nodes). Our study also proposes a unique application of the coarsening algorithm for graph-level tasks, including graph classification and graph regression, which have not yet been explored. We conduct extensive experiments on multiple benchmark datasets in the order of $100K$ nodes to evaluate the performance of our approach. The results demonstrate that our method achieves competitive performance in tasks involving classification and regression on nodes and graphs, compared to traditional GNNs, while having single-node inference times that are orders of magnitude faster. Furthermore, our approach significantly reduces memory consumption, allowing training and inference on low-resource devices where traditional methods struggle.
Code
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
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Graph Classification | AIDS | FIT-GNN | Accuracy | 84.3 | #1 of 2 | Archive leaderboard | report |
| Graph Classification | AIDS | FIT-GNN | Inference Time (ms) | 0.00155 | #1 of 2 | Archive leaderboard | report |
| Graph Classification | PROTEINS | FIT-GNN | Accuracy | 82.1 | #5 of 103 | Archive leaderboard | report |
| Graph Classification | PROTEINS | FIT-GNN | Inference Time (ms) | 0.0016 | #5 of 103 | Archive leaderboard | report |
| Graph Regression | QM9 | FIT-GNN | Inference Time (ms) | 0.0018 | #1 of 1 | Archive leaderboard | report |
| Graph Regression | QM9: UATOM | FIT-GNN | MAE | 0.63 | #1 of 1 | Archive leaderboard | report |
| Graph Regression | QM9: ZPVE | FIT-GNN | MAE | 0.818 | #1 of 1 | Archive leaderboard | report |
| Graph Regression | QM9: del e | FIT-GNN | MAE | 0.875 | #1 of 1 | Archive leaderboard | report |
| Graph Regression | QM9: mu | FIT-GNN | MAE | 0.841 | #1 of 1 | Archive leaderboard | report |
| Graph Regression | ZINC 10k | FIT-GNN | Inference Time (ms) | 0.00184 | #1 of 1 | Archive leaderboard | report |
| Graph Regression | ZINC 10k | FIT-GNN | MAE | 0.578 | #1 of 1 | Archive leaderboard | report |
| Node Classification | CiteSeer: 5 nodes per class | FIT-GNN | Accuracy | 62.4 | #1 of 1 | Archive leaderboard | report |
| Node Classification | Citeseer | FIT-GNN | Inference Time (ms) | 0.0018 | #71 of 71 | Archive leaderboard | report |
| Node Classification | Coauthor CS | FIT-GNN | Inference Time (ms) | 0.0017 | #24 of 24 | Archive leaderboard | report |
| Node Classification | Cora | FIT-GNN | Accuracy | 82.9% | #48 of 73 | Archive leaderboard | report |
| Node Classification | Cora | FIT-GNN | Inference Time (ms) | 0.0019 | #48 of 73 | Archive leaderboard | report |
| Node Classification | Cora: 5 nodes per class | FIT-GNN | Accuracy | 72.9 | #1 of 1 | Archive leaderboard | report |
| Node Classification | DBLP | FIT-GNN | Inference Time (ms) | 0.0018 | #6 of 6 | Archive leaderboard | report |
| Node Classification | DBLP: 20 nodes per class | FIT-GNN | Accuracy | 0.789 | #1 of 1 | Archive leaderboard | report |
| Node Classification | DBLP: 5 nodes per class | FIT-GNN | Accuracy | 68.3 | #1 of 1 | Archive leaderboard | report |
| Node Classification | PubMed: 5 nodes per class | FIT-GNN | Accuracy | 67.6 | #1 of 1 | Archive leaderboard | report |
| Node Classification | Pubmed | FIT-GNN | Inference Time (ms) | 0.0018 | #70 of 70 | Archive leaderboard | report |
| Node Classification | ogbn-products | FIT-GNN | Inference Time (ms) | 0.0016 | #1 of 1 | Archive leaderboard | report |
| Node Classification | ogbn-products: 20 nodes per class | FIT-GNN | Accuracy | 0.406 | #1 of 1 | 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