Table 3. Open in new tab Performance of... | Oxford Academic

Table 3.

Open in new tab

Performance of different receptive fields (RMSE).

Receptive field	GCN	GraphSAGE	GAT	GraphTransformer	Impeller
2	0.339 ± 0.000	0.352 ± 0.008	0.360 ± 0.005	1.058 ± 0.463	$0.310 \pm 0.016$
4	0.348 ± 0.001	0.362 ± 0.013	0.372 ± 0.022	0.424 ± 0.031	$0.286 \pm 0.009$
8	0.386 ± 0.012	0.496 ± 0.033	0.454 ± 0.024	0.351 ± 0.002	$0.279 \pm 0.001$
16	0.403 ± 0.001	0.617 ± 0.054	0.506 ± 0.061	0.435 ± 0.017	$0.286 \pm 0.010$
32	0.418 ± 0.015	1.466 ± 0.031	1.458 ± 0.037	0.420 ± 0.000	$0.277 \pm 0.002$
64	0.430 ± 0.002	1.615 ± 0.010	1.621 ± 0.004	0.420 ± 0.001	$0.302 \pm 0.028$
128	0.429 ± 0.001	1.629 ± 0.003	1.614 ± 0.022	0.420 ± 0.001	$0.357 \pm 0.001$

Receptive field	GCN	GraphSAGE	GAT	GraphTransformer	Impeller
2	0.339 ± 0.000	0.352 ± 0.008	0.360 ± 0.005	1.058 ± 0.463	$0.310 \pm 0.016$
4	0.348 ± 0.001	0.362 ± 0.013	0.372 ± 0.022	0.424 ± 0.031	$0.286 \pm 0.009$
8	0.386 ± 0.012	0.496 ± 0.033	0.454 ± 0.024	0.351 ± 0.002	$0.279 \pm 0.001$
16	0.403 ± 0.001	0.617 ± 0.054	0.506 ± 0.061	0.435 ± 0.017	$0.286 \pm 0.010$
32	0.418 ± 0.015	1.466 ± 0.031	1.458 ± 0.037	0.420 ± 0.000	$0.277 \pm 0.002$
64	0.430 ± 0.002	1.615 ± 0.010	1.621 ± 0.004	0.420 ± 0.001	$0.302 \pm 0.028$
128	0.429 ± 0.001	1.629 ± 0.003	1.614 ± 0.022	0.420 ± 0.001	$0.357 \pm 0.001$

The best imputation performance is highlighted in bold.

Table 3.

Open in new tab

Performance of different receptive fields (RMSE).

Receptive field	GCN	GraphSAGE	GAT	GraphTransformer	Impeller
2	0.339 ± 0.000	0.352 ± 0.008	0.360 ± 0.005	1.058 ± 0.463	$0.310 \pm 0.016$
4	0.348 ± 0.001	0.362 ± 0.013	0.372 ± 0.022	0.424 ± 0.031	$0.286 \pm 0.009$
8	0.386 ± 0.012	0.496 ± 0.033	0.454 ± 0.024	0.351 ± 0.002	$0.279 \pm 0.001$
16	0.403 ± 0.001	0.617 ± 0.054	0.506 ± 0.061	0.435 ± 0.017	$0.286 \pm 0.010$
32	0.418 ± 0.015	1.466 ± 0.031	1.458 ± 0.037	0.420 ± 0.000	$0.277 \pm 0.002$
64	0.430 ± 0.002	1.615 ± 0.010	1.621 ± 0.004	0.420 ± 0.001	$0.302 \pm 0.028$
128	0.429 ± 0.001	1.629 ± 0.003	1.614 ± 0.022	0.420 ± 0.001	$0.357 \pm 0.001$

Receptive field	GCN	GraphSAGE	GAT	GraphTransformer	Impeller
2	0.339 ± 0.000	0.352 ± 0.008	0.360 ± 0.005	1.058 ± 0.463	$0.310 \pm 0.016$
4	0.348 ± 0.001	0.362 ± 0.013	0.372 ± 0.022	0.424 ± 0.031	$0.286 \pm 0.009$
8	0.386 ± 0.012	0.496 ± 0.033	0.454 ± 0.024	0.351 ± 0.002	$0.279 \pm 0.001$
16	0.403 ± 0.001	0.617 ± 0.054	0.506 ± 0.061	0.435 ± 0.017	$0.286 \pm 0.010$
32	0.418 ± 0.015	1.466 ± 0.031	1.458 ± 0.037	0.420 ± 0.000	$0.277 \pm 0.002$
64	0.430 ± 0.002	1.615 ± 0.010	1.621 ± 0.004	0.420 ± 0.001	$0.302 \pm 0.028$
128	0.429 ± 0.001	1.629 ± 0.003	1.614 ± 0.022	0.420 ± 0.001	$0.357 \pm 0.001$

The best imputation performance is highlighted in bold.