Graph Regression on QM9 (test)
0.231muPPGN
Evaluation results
| Method | Links | ||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PPGNComplexity=O(T|V|^3)2023.10 | 0.231 | 0.382 | 0.0028 | 0.0029 | 0.0041 | 16.07 | 0.0064 | 0.234 | 0.184 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.234 | 0.229 | 0.238 | |
| MAG-GNNComplexity=O(m_t T |V|^2)2023.10 | 0.353 | 0.226 | 0.0026 | 0.0025 | 0.0035 | 15.44 | 0.0002 | 0.111 | 0.093 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.105 | 0.089 | 0.116 | |
| ARMAArchitecture=GNN(64)-AvgPool-FC(128), Train-Val-Test Split=80-10-102019.01 | 0.394 | 0.098 | 0.326 | 0.508 | 0.552 | 0.119 | 0.338 | 0.053 | 0.163 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RNMComplexity=O(kT|V|^2)2023.10 | 0.426 | 0.306 | 0.0026 | 0.0027 | 0.0047 | 20.9 | 0.0002 | 0.281 | 0.177 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.193 | 0.384 | 0.25 | |
| NGNNComplexity=O(T|V|^3)2023.10 | 0.428 | 0.23 | 0.0027 | 0.003 | 0.0038 | 20.5 | 0.0002 | 0.295 | 0.174 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.361 | 0.305 | 0.489 | |
| I2-GNNComplexity=O(T|V|^4)2023.10 | 0.428 | 0.23 | 0.0026 | 0.0027 | 0.0038 | 18.64 | 0.0001 | 0.211 | 0.073 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.206 | 0.269 | 0.261 | |
| ChebyshevArchitecture=GNN(64)-AvgPool-FC(128), Train-Val-Test Split=80-10-102019.01 | 0.433 | 0.171 | 0.391 | 0.528 | 0.565 | 0.294 | 0.358 | 0.126 | 0.215 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CayleyNetArchitecture=GNN(64)-AvgPool-FC(128), Train-Val-Test Split=80-10-102019.01 | 0.442 | 0.118 | 0.336 | 0.679 | 0.758 | 0.185 | 0.555 | 1.493 | 0.184 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GCNArchitecture=GNN(64)-AvgPool-FC(128), Train-Val-Test Split=80-10-102019.01 | 0.445 | 0.141 | 0.371 | 0.584 | 0.65 | 0.132 | 0.349 | 0.064 | 0.192 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 1-2-3-GNNComplexity=O(T|V|^4)2023.10 | 0.476 | 0.27 | 0.0034 | 0.0035 | 0.0048 | 22.9 | 0.0002 | 0.0427 | 0.0944 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.111 | 0.0419 | 0.0469 | |
| 1-2-3 GNN2021.11 | — | — | — | — | — | — | — | — | — | 0.473 | 0.27 | 0.0034 | 0.0035 | 0.0048 | 22.9 | 0.0002 | 0.0427 | 0.111 | 0.0419 | 0.0469 | 0.0944 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 1-GNN2021.11 | — | — | — | — | — | — | — | — | — | 0.493 | 0.78 | 0.0032 | 0.0035 | 0.0049 | 34.1 | 0.0012 | 2.32 | 2.08 | 2.23 | 1.94 | 0.27 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CAGradDescription=Conflict-Averse Gradient Descent2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 5.7 | 112.8 | — | — | — | |
| Drop-1-GNNbase_model=1-GNN, epochs=3002021.11 | — | — | — | — | — | — | — | — | — | 0.453 | 0.767 | 0.0031 | 0.0031 | 0.0046 | 30.83 | 0.0009 | 1.8 | 1.86 | 2 | 2.12 | 0.259 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DropMPNNbase_model=MPNN, epochs=3002021.11 | — | — | — | — | — | — | — | — | — | 0.059 | 0.173 | 0.0016 | 0.0018 | 0.0028 | 0.392 | 0.0001 | 0.0409 | 0.0536 | 0.0481 | 0.0508 | 0.0396 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DWADescription=Dynamic Weight Averaging2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 6.4 | 175.3 | — | — | — | |
| GAT2021.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 2.68 | 4.65 | 1.48 | 1.53 | 2.31 | 52.39 | 14.87 | 7.61 | 6.86 | 7.64 | 6.54 | 4.11 | 1.48 | 0 | — | — | — | — | — | |
| GATv22021.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 2.65 | 4.28 | 1.41 | 1.47 | 2.29 | 16.37 | 14.03 | 6.07 | 6.28 | 6.6 | 5.97 | 3.57 | 1.59 | -11.5 | — | — | — | — | — | |
| GCN2021.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 3.21 | 4.22 | 1.45 | 1.62 | 2.42 | 16.38 | 17.4 | 7.82 | 8.24 | 9.05 | 7 | 3.93 | 1.02 | -1.5 | — | — | — | — | — | |
| GIN2021.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 2.64 | 4.67 | 1.42 | 1.5 | 2.27 | 15.63 | 12.93 | 5.88 | 18.71 | 5.62 | 5.38 | 3.53 | 1.05 | -2.3 | — | — | — | — | — | |
| IMTL-GDescription=Independent Multi-Task Learning - Gradient2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 4.7 | 77.2 | — | — | — | |
| LSDescription=Linear Scalarization2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 6.8 | 177.6 | — | — | — | |
| MGDADescription=Multiple-Gradient Descent Algorithm2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 5.9 | 120.5 | — | — | — | |
| MPNN2021.11 | — | — | — | — | — | — | — | — | — | 0.358 | 0.89 | 0.0054 | 0.0062 | 0.0066 | 28.5 | 0.0022 | 2.05 | 2 | 2.02 | 2.02 | 0.42 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Nash-MTL2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 2.5 | 62 | — | — | — | |
| PCGradDescription=Projected Conflicting Gradients2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 5 | 125.7 | — | — | — | |
| PPGN2021.11 | — | — | — | — | — | — | — | — | — | 0.0934 | 0.318 | 0.0017 | 0.0021 | 0.0029 | 3.78 | 0.0004 | 0.022 | 0.0504 | 0.0294 | 0.24 | 0.0144 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RLWDescription=Random Loss Weighting2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 8.2 | 203.8 | — | — | — | |
| SIDescription=Scale Invariant2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 4 | 77.8 | — | — | — | |
| UWDescription=Uncertainty Weighting2022.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 5.3 | 108 | — | — | — |