Semi-supervised underwater acoustic source localization based on residual convolutional autoencoder

EURASIP Journal on Advances in Signal Processing

Table 2 Localization performance of narrowband dataset

	Percentage of label data	75%		37.5%		15%
Frequency points	Positioning network model	MAE (km)	\(P_{CL - 5\% }\)_(%)	MAE (km)	\(P_{CL - 5\% }\)_(%)	MAE (km)	\(P_{CL - 5\% }\)_(%)
109	RA-CAE-SSL	0.9098	62	1.2392	55	1.5352	44.5
	CAE-SSL	0.9072	63.5	1.4963	50.5	1.7699	43.5
	RA-SL	0.8825	65	1.4024	51.5	1.8392	41.5
	CNN	0.7986	61.5	1.3297	46	1.9254	39.5
127	RA-CAE-SSL	0.6581	75.50	0.7100	72	0.9487	64
	CAE-SSL	0.6719	72.5	0.6489	72	1.2943	55.5
	RA-SL	0.7328	73	0.8213	67	1.1592	57.5
	CNN	0.7812	68	0.7570	66.50	1.2157	54.50
145	RA-CAE-SSL	0.3606	86	0.5211	77	0.6323	73
	CAE-SSL	0.4503	83	0.5954	76	0.6746	71
	RA-SL	0.4761	81	0.6873	73.5	0.8090	72
	CNN	0.4437	81.5	0.6897	72	0.7254	69
163	RA-CAE-SSL	0.1201	93.5	0.2259	89	0.4083	81
	CAE-SSL	0.1390	93	0.2679	89.5	0.5134	79
	RA-SL	0.1570	92	0.3891	86	0.5954	79.5
	CNN	0.1255	91.5	0.3094	86	0.7659	68
198	RA-CAE-SSL	0.1470	95.5	0.2802	86.5	0.4557	82
	CAE-SSL	0.1155	95.5	0.2090	90	0.4372	82
	RA-SL	0.1489	94	0.2428	88	0.5449	80
	CNN	0.1205	94	0.2413	87.5	0.5700	78