Graph Star Net for Generalized Multi-Task Learning

21 Jun 2019Lu HaonanSeth H. HuangTian YeGuo Xiuyan

In this work, we present graph star net (GraphStar), a novel and unified graph neural net architecture which utilizes message-passing relay and attention mechanism for multiple prediction tasks - node classification, graph classification and link prediction. GraphStar addresses many earlier challenges facing graph neural nets and achieves non-local representation without increasing the model depth or bearing heavy computational costs... (read more)

PDF Abstract

Evaluation results from the paper


Task Dataset Model Metric name Metric value Global rank Compare
Text Classification 20NEWS GraphStar Accuracy 86.9 # 2
Node Classification Citeseer GraphStar Accuracy 71.00% # 12
Link Prediction Citeseer GraphStar Accuracy 97.70% # 1
Link Prediction Citeseer GraphStar AUC 97.47 # 1
Link Prediction Citeseer GraphStar AP 97.93 # 1
Node Classification Cora GraphStar Accuracy 82.1% # 10
Link Prediction Cora GraphStar Accuracy 95.90% # 1
Link Prediction Cora GraphStar AUC 95.65 # 1
Link Prediction Cora GraphStar AP 96.15 # 1
Graph Classification D&D GraphStar Accuracy 79.60% # 2
Sentiment Analysis IMDb GraphStar Accuracy 96.0 # 2
Sentiment Analysis MR GraphStar Accuracy 76.6 # 9
Graph Classification MUTAG GraphStar Accuracy 91.2% # 1
Text Classification Ohsumed GraphStar Accuracy 64.2 # 3
Node Classification PPI GraphStar F1 99.40% # 1
Graph Classification PROTEINS GraphStar Accuracy 77.90% # 1
Node Classification Pubmed GraphStar Accuracy 77.2% # 10
Link Prediction Pubmed GraphStar Accuracy 97.00% # 1
Link Prediction Pubmed GraphStar AUC 97.10 # 2
Link Prediction Pubmed GraphStar AP 96.90 # 2
Text Classification R52 GraphStar Accuracy 95.00 # 1
Text Classification R8 GraphStar Accuracy 97.4 # 1