DropGNN: Random Dropouts Increase the Expressiveness of Graph Neural Networks

NeurIPS 2021  ·  Pál András Papp, Karolis Martinkus, Lukas Faber, Roger Wattenhofer ·

This paper studies Dropout Graph Neural Networks (DropGNNs), a new approach that aims to overcome the limitations of standard GNN frameworks. In DropGNNs, we execute multiple runs of a GNN on the input graph, with some of the nodes randomly and independently dropped in each of these runs. Then, we combine the results of these runs to obtain the final result. We prove that DropGNNs can distinguish various graph neighborhoods that cannot be separated by message passing GNNs. We derive theoretical bounds for the number of runs required to ensure a reliable distribution of dropouts, and we prove several properties regarding the expressive capabilities and limits of DropGNNs. We experimentally validate our theoretical findings on expressiveness. Furthermore, we show that DropGNNs perform competitively on established GNN benchmarks.

PDF Abstract NeurIPS 2021 PDF NeurIPS 2021 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Graph Classification D&D DropGIN Accuracy 78.151±3.711 # 29
Compare
Graph Classification ENZYMES DropGIN Accuracy 65.128±4.117 # 22
Compare
Molecular Property Prediction ESOL DropGIN RMSE 0.520±0.048 # 4
Compare
R2 0.935±0.012 # 5
Compare
Graph Regression ESR2 DropGIN R2 0.675±0.000 # 3
Compare
RMSE 0.503±0.675 # 3
Compare
Graph Regression F2 DropGIN R2 0.886±0.000 # 4
Compare
RMSE 0.343±0.886 # 4
Compare
Molecular Property Prediction FreeSolv DropGIN RMSE 0.657±0.059 # 2
Compare
R2 0.972±0.005 # 2
Compare
Graph Classification IMDb-B DropGIN Accuracy 75.7% # 20
Compare
Graph Classification IMDb-M DropGIN Accuracy 51.4% # 15
Compare
Graph Regression KIT GINDrop R2 0.835±0.000 # 3
Compare
RMSE 0.441±0.835 # 3
Compare
Graph Regression Lipophilicity DropGIN RMSE 0.552±0.012 # 5
Compare
R2 0.809±0.008 # 5
Compare
Graph Classification MUTAG DropGIN Accuracy 90.4% # 21
Compare
Graph Classification NCI1 DropGIN Accuracy 84.331±1.564 # 21
Compare
Graph Classification NCI109 DropGIN Accuracy 83.961±1.141 # 8
Compare
Graph Regression PARP1 DropGIN R2 0.920±0.000 # 5
Compare
RMSE 0.354±0.920 # 5
Compare
Graph Regression PGR GINDrop R2 0.702±0.000 # 3
Compare
RMSE 0.527±0.702 # 3
Compare
Graph Classification PROTEINS DropGIN Accuracy 76.3% # 52
Compare
Graph Classification PTC DropGIN Accuracy 66.3% # 19
Compare

Methods
Add Remove

Contact us on: hello@paperswithcode.com . Papers With Code is a free resource with all data licensed under CC-BY-SA.
Terms Data policy Cookies policy