Node Classification On Arxiv Year

Metrics

Accuracy

Results

Performance results of various models on this benchmark

		Paper Title
FaberNet	64.62±1.01	HoloNets: Spectral Convolutions do extend to Directed Graphs
Dir-GNN	64.08±0.26	Edge Directionality Improves Learning on Heterophilic Graphs
G^2-GraphSAGE	63.30±1.84	Gradient Gating for Deep Multi-Rate Learning on Graphs
Dual-Net GNN	62.65±0.39	Feature Selection: Key to Enhance Node Classification with Graph Neural Networks
OGNN	56.50±0.13	Uplifting Message Passing Neural Network with Graph Original Information
LINKX	56.00±1.34	Large Scale Learning on Non-Homophilous Graphs: New Benchmarks and Strong Simple Methods
LW-GCN	55.8±0.2	Label-Wise Graph Convolutional Network for Heterophilic Graphs
GloGNN++	54.79±0.25	Finding Global Homophily in Graph Neural Networks When Meeting Heterophily
IIE-GNN	51.59±0.24	Enhancing Intra-class Information Extraction for Heterophilous Graphs: One Neural Architecture Search Approach
DJ-GNN	49.21±0.20	Diffusion-Jump GNNs: Homophiliation via Learnable Metric Filters
GESN	48.80 ± 0.22	Addressing Heterophily in Node Classification with Graph Echo State Networks

0 of 11 row(s) selected.