Table 6 Cross-domain graph classification on Letter-Drawing dataset.
From: TO-UGDA: target-oriented unsupervised graph domain adaptation
Methods | Low\(\rightarrow\)Med | Low\(\rightarrow\)High | Med\(\rightarrow\)Low | Med\(\rightarrow\)High | High\(\rightarrow\)Low | High\(\rightarrow\)Med | Avg. |
|---|---|---|---|---|---|---|---|
GCN59 | 0.4822 | 0.2982 | 0.6382 | 0.2800 | 0.5391 | 0.3884 | 0.4377 |
SGC60 | 0.5253 | 0.2160 | 0.7921 | 0.2908 | 0.5346 | 0.3986 | 0.4596 |
GIN44 | 0.4973 | 0.2102 | 0.7812 | 0.2849 | 0.5049 | 0.4102 | 0.4481 |
CDAN35 | 0.4012 | 0.2367 | 0.6587 | 0.2154 | 0.5071 | 0.4483 | 0.4112 |
ToAlign65 | 0.5829 | 0.2570 | 0.6374 | 0.2714 | 0.5946 | 0.4768 | 0.4700 |
MetaAlign66 | 0.5271 | 0.2638 | 0.6819 | 0.2932 | 0.5623 | 0.5122 | 0.4734 |
DEAL24 | 0.5731 | 0.3189 | 0.7413 | 0.2911 | 0.6136 | 0.5587 | 0.5161 |
COCO48 | 0.5965 | 0.3347 | 0.7893 | 0.3149 | 0.7445 | 0.5947 | 0.5624 |
TO-UGDA | 0.6212 | 0.3473 | 0.8526 | 0.3355 | 0.8318 | 0.6139 | 0.6004 |