Graph Classification On Imdb B

Accuracy

Results

Performance results of various models on this benchmark

		Paper Title
U2GNN (Unsupervised)	96.41%	Universal Graph Transformer Self-Attention Networks
ESA (Edge set attention, no positional encodings)	86.250±0.957	An end-to-end attention-based approach for learning on graphs
GAT	84.250±2.062	Graph Attention Networks
MEWISPool	82.13%	Maximum Entropy Weighted Independent Set Pooling for Graph Neural Networks
GIN	81.250±3.775	How Powerful are Graph Neural Networks?
TokenGT	80.250±3.304	Pure Transformers are Powerful Graph Learners
GATv2	80.000±2.739	How Attentive are Graph Attention Networks?
G_ResNet	79.90%	When Work Matters: Transforming Classical Network Structures to Graph CNN
GCN	79.500±3.109	Semi-Supervised Classification with Graph Convolutional Networks
GraphGPS	79.250±3.096	Recipe for a General, Powerful, Scalable Graph Transformer
DUGNN	78.70%	Learning Universal Graph Neural Network Embeddings With Aid Of Transfer Learning
TFGW ADJ (L=2)	78.3%	Template based Graph Neural Network with Optimal Transport Distances
PNA	78.000±3.808	Principal Neighbourhood Aggregation for Graph Nets
sGIN	77.94%	Mutual Information Maximization in Graph Neural Networks
Graphormer	77.500±2.646	Do Transformers Really Perform Bad for Graph Representation?
SEG-BERT	77.2%	Segmented Graph-Bert for Graph Instance Modeling
U2GNN	77.04%	Universal Graph Transformer Self-Attention Networks
PIN	76.6%	Weisfeiler and Lehman Go Paths: Learning Topological Features via Path Complexes
GIUNet	76%	Graph isomorphism UNet
DropGIN	75.7%	DropGNN: Random Dropouts Increase the Expressiveness of Graph Neural Networks

0 of 50 row(s) selected.