Recent Progress of Anomaly Detection

Table 2

Experimental datasets used in our experiments.

Dataset	N(A)	Attribute	Anomalies	Source
ALOI	50000(1508)	27	The 1508 of rare objects	UCI [53]
Arcene	100(44)	10000	The cancer patients	UCI [53]
Ionoshere	351(126)	32	The ‘bad’ class	UCI [53]
KDDCup99	48113(200)	38	The U2R class	UCI [53]
PenDigits	9868(20)	16	The fourth class	UCI [53]
Sonor	208(97)	60	The rock object	UCI [53]
WDBC	367(10)	30	The malignant class	UCI [53]
Waveform	3443(100)	21	The ‘0’ class	UCI [53]
Ann-thyroid	7129(534)	21	The hyper and subnormal classes	ELKI [54]
Arrhythmia	450(206)	259	The arrhythmia class	ELKI [54]
HeartDisease	270(120)	13	The affected patients class	ELKI [54]
Pima	768(268)	8	The Diabetes class	ELKI [54]
SpamBase	4209(1681)	57	The non-spam class	ELKI [54]
ALLAML	38(11)	7129	The AML class	EBD [56]
DLBCL	77(19)	7129	The FL morphology class	EBD [56]
Gisette	550(50)	5000	The normal class	EBD [56]
Lung_MPM	181(31)	12533	The MPM class	EBD [56]
Ovarian	253(162)	15154	The Ovarian Cancer class	EBD [56]