TABLE I. Classification Performances of the Proposed CSEN and Competing Methods. The Best COVID-19 Recognition Rates Are Highlighted.

		k-NN	SVM	MLP	CRC	ReconNet	CSEN1	CSEN2
Accuracy	Bacterial	0.777	0.780	0.763	0.820	0.765	0.793	0.794
	Viral	0.801	0.787	0.765	0.827	0.785	0.805	0.803
	Normal	0.903	0.934	0.933	0.928	0.918	0.926	0.927
	COVID-19	0.950	0.945	0.949	0.955	0.936	0.955	0.959
TN	Bacterial	3166	3219	3114	3063	3180	3177	3173
	Viral	4123	3965	3923	4385	4005	4109	4091
	Normal	4253	4444	4442	4380	4364	4388	4396
	COVID-19	5525	5489	5522	5554	5435	5548	5572
TP	Bacterial	1720	1687	1680	2091	1629	1810	1818
	Viral	909	979	884	816	928	954	959
	Normal	1420	1427	1421	1456	1407	1431	1428
	COVID-19	446	452	444	447	448	455	455
FP	Bacterial	360	307	412	463	346	349	353
	Viral	678	836	878	416	796	692	710
	Normal	454	263	265	327	343	319	311
	COVID-19	299	335	302	270	389	276	252
FN	Bacterial	1040	1073	1080	669	1131	950	942
	Viral	576	506	601	669	557	531	526
	Normal	159	152	158	123	172	148	151
	COVID-19	16	10	18	15	14	7	7
Sensitivity	Bacterial	0.623	0.611	0.609	0.758	0.590	0.656	0.659
	Viral	0.612	0.660	0.595	0.550	0.625	0.642	0.646
	Normal	0.899	0.904	0.900	0.922	0.891	0.906	0.904
	COVID-19	0.965	0.978	0.961	0.968	0.970	0.985	0.985
Specificity	Bacterial	0.898	0.913	0.883	0.869	0.902	0.901	0.900
	Viral	0.859	0.826	0.817	0.913	0.834	0.856	.852
	Normal	0.904	0.944	0.944	0.931	0.927	0.932	0.934
	COVID-19	0.949	0.943	0.948	0.954	0.933	0.953	0.957
F1-score	Bacterial	0.711	0.710	0.693	0.787	0.688	0.736	0.737
	Viral	0.592	0.593	0.545	0.601	0.578	0.609	0.608
	Normal	0.823	0.873	0.870	0.866	0.845	0.860	0.861
	COVID-19	0.740	0.724	0.735	0.758	0.690	0.763	0.778