Table 3.

Pairwise comparison for the average positioning error and convergence speed using “Tukey-Kramer” for a 95% confidence interval on a two-way ANOVA.

	Group 1	Group 2	Lower	Mean diff.	Upper	p-Value	Mean diff (%)
Performance	R	R + P	−0.0121	0.0028	0.0177	0.9627	7.18
	R	R + N	−0.0027	0.0122	0.0271	0.1523	31.28
	R	R + P + N	−0.0114	0.0036	0.0185	0.9278	9.23
	R + P	R + N	−0.0055	0.0094	0.0243	0.3681	25.97
	R + P	R + P + N	−0.0142	0.0008	0.0157	0.9992	2.21
	R + N	R + P + N	−0.0236	−0.0086	0.0063	0.4442	−32.09
Speed	R	R + P	−0.7232	−0.5180	−0.3129	0.0000	−44.29
	R	R + N	−0.1879	0.0173	0.2224	0.9964	1.48
	R	R + P + N	−0.4286	−0.2234	−0.0183	0.0264	−19.10
	R + P	R + N	0.3302	0.5353	0.7405	0.0000	31.72
	R + P	R + P + N	0.0894	0.2946	0.4997	0.0013	17.46
	R + N	R + P + N	−0.4459	−0.2407	−0.0356	0.0137	−20.89

Here, we analyze the average performance and convergence speed on the validation set of the best 4 hyperparameter sets for each condition. All hyperparameter sets were tested with 10 different randomly initialized networks. A positive percentage indicates the effect size in which Group 2 improves upon Group 1, whereas a negative percentage indicates the effect size in which Group 2 is worse than Group 1.