Table 6.

Result table with randomly initialized neural network on PI.

# layers	# hidden units	Sens.	Spec.	F1	Mac. Ave.	Mic. Ave.
1	512	0.769	0.903	0.692	0.836	0.879
1	1024	0.788	0.897	0.695	0.843	0.878
1	1536	0.805	0.889	0.693	0.847	0.875
1	2048	0.803	0.891	0.694	0.847	0.876
2	512	0.741	0.909	0.684	0.825	0.88
2	1024	0.765	0.902	0.688	0.833	0.878
2	1536	0.747	0.905	0.682	0.826	0.877
2	2048	0.77	0.904	0.693	0.837	0.88
3	512	0.763	0.897	0.68	0.83	0.874
3	1024	0.755	0.908	0.691	0.832	0.881
3	1536	0.757	0.905	0.688	0.831	0.879
3	2048	0.753	0.904	0.683	0.828	0.877
4	512	0.765	0.886	0.666	0.826	0.865
4	1024	0.759	0.9	0.681	0.829	0.875
4	1536	0.771	0.893	0.679	0.832	0.872
4	2048	0.709	0.894	0.642	0.801	0.861