TABLE 5.

Experiment 2 - detailed metrics of the best run of each algorithm.

Metric	RNN-HASCA	RNN-SCA	RNN-GA	RNN-PSO	RNN-FA	RNN-BSO	RNN-RSA	RNN-COLSHADE
Accuracy (%)	96.563	31.3830	85.3519	89.8936	77.5777	77.5777	86.7840	91.2848
Precision Normal	0.902148	0.158545	0.521622	0.862155	0.000000	0.000000	0.874439	0.856492
Precision Anomaly	0.830601	0.069825	0.000000	0.420195	0.000000	0.000000	0.348780	0.492537
Precision Seizure	0.993485	0.779940	0.997653	0.991945	0.775777	0.775777	1.000000	0.991940
W.Avg. Precision	0.968380	0.634874	0.857194	0.934271	0.601831	0.601831	0.937864	0.938041
Recall Normal	0.969231	0.558974	0.989744	0.882051	0.000000	0.000000	1.000000	0.964103
Recall Anomaly	0.962025	0.177215	0.000000	0.816456	0.000000	0.000000	0.905063	0.835443
Recall Seizure	0.965190	0.274789	0.896624	0.909283	1.000000	1.000000	0.837553	0.908755
W.Avg. Recall	0.965630	0.31383	0.853519	0.898936	0.775777	0.775777	0.867840	0.912848
F1-score Normal	0.969231	0.558974	0.989744	0.882051	0.000000	0.000000	1.000000	0.964103
F1-score Anomaly	0.962025	0.177215	0.000000	0.816456	0.000000	0.000000	0.905063	0.835443
F1-score Seizure	0.965190	0.274789	0.896624	0.909283	1.000000	1.000000	0.837553	0.908755
W.Avg. F1-score	0.965630	0.31383	0.853519	0.898936	0.775777	0.775777	0.867840	0.912848

In all tables that contain simulation results, the best score in every category is marked bold.