Towards a deep learning-based outlier detection approach in the context of streaming data

Journal of Big Data

Table 4 Performance evaluation of different DL methods (%)

Dataset	Algorithm	Accuracy	TNR	FAR	FNR	Precision	Recall	F1-score
Breast-Cancer	DeepAnT [43]	93.618	94.361	5.639	6.918	94.067	93.082	93.541
	RobustTAD [44]	95.864	96.938	3.062	4.789	96.243	95.211	95.642
	Proposed model	98.354	96.544	3.456	2.215	98.579	97.785	98.108
Annthyroid	DeepAnT [43]	91.037	92.304	7.696	9.164	91.210	90.836	91.120
	RobustTAD [44]	95.646	96.408	3.592	4.543	95.914	95.457	95.603
	Proposed model	98.644	97.034	2.966	1.706	98.853	98.294	98.543
Musk	DeepAnT [43]	90.392	92.143	7.857	10.195	90.491	89.805	90.098
	RobustTAD [44]	94.451	93.469	6.531	5.979	94.670	94.021	94.392
	Proposed model	99.241	97.493	2.507	1.268	98.903	98.732	98.985
Cardiotocography	DeepAnT [43]	90.816	91.362	8.638	10.368	90.984	89.632	90.620
	RobustTAD [44]	93.828	92.713	7.287	6.025	95.017	93.975	94.749
	Proposed model	99.631	97.682	2.318	0.986	99.702	99.014	99.437