Table 3.

Class-wise performance of the ESN and RF on the fixed test dataset ($n=321$) in experiment 4

		PRC			REC			SPC			F1-score
		${\mathit{\phi }}_0=0$	rnd. ${\mathit{\phi }}_0$	$|\mathrm{\Delta }|$	${\mathit{\phi }}_0=0$	rnd. ${\mathit{\phi }}_0$	$|\mathrm{\Delta }|$	${\mathit{\phi }}_0=0$	rnd. ${\mathit{\phi }}_0$	$|\mathrm{\Delta }|$	${\mathit{\phi }}_0=0$	rnd. ${\mathit{\phi }}_0$	$|\mathrm{\Delta }|$
$C_1$	ESN	0.791	0.791	0.000	0.928	0.928	0.000	0.934	0.934	0.000	0.853	0.853	0.000
$n_1=69$	RF	0.903	0.868	0.035	0.942	0.957	0.015	0.972	0.960	0.012	0.922	0.910	0.012
$C_2$	ESN	0.897	0.897	0.000	0.625	0.625	0.000	0.985	0.985	0.000	0.737	0.737	0.000
$n_2=56$	RF	0.946	0.939	0.007	0.625	0.554	0.071	0.992	0.992	0.000	0.753	0.697	0.056
$C_3$	ESN	0.929	0.912	0.017	0.853	0.853	0.000	0.985	0.985	0.000	0.889	0.881	0.008
$n_3=61$	RF	0.889	0.803	0.086	0.918	0.869	0.049	0.973	0.950	0.023	0.903	0.835	0.068
$C_4$	ESN	0.853	0.865	0.012	0.985	0.985	0.000	0.957	0.961	0.004	0.914	0.921	0.007
$n_4=65$	RF	0.785	0.763	0.022	0.954	0.892	0.062	0.934	0.930	0.004	0.861	0.823	0.038
$C_{\mathrm{bg}}$	ESN	1.000	1.000	0.000	1.000	1.000	0.000	1.000	1.000	0.000	1.000	1.000	0.000
$n_{\mathrm{bg}}=69$	RF	1.000	1.000	0.000	1.000	1.000	0.000	1.000	1.000	0.000	1.000	1.000	0.000
w.m.	ESN	0.894	0.893	0.001	0.888	0.888	0.000	0.971	0.971	0.000	0.884	0.884	0.000
$n=321$	RF	0.905	0.876	0.029	0.897	0.866	0.031	0.974	0.966	0.008	0.894	0.860	0.034

The last row contains the weighted mean (w.m.) of the measures according to the class distribution in the test set. The differences between the performance at the default ($\mathit{\phi }=0$) and the random starting angle (rnd. ${\mathit{\phi }}_0$) are computed as absolute differences $|\mathrm{\Delta }|$. The ESN recognized the same cells under different rotation angles more constantly. Superior results are printed in bold. The best hyper-parameters of the classifiers were chosen according to the best cross-validation results of experiment 2 (ESN: $N=4000$, RF: forest size $=128$, maximum tree depth $=12$)