Table 1.

Description of datasets.

Dataset Name and Source	No. Observations	No. Features	No. Classes	Dimensionality *
Jasmine ${}^1$	2984 (1492/1492)	145	2	0.048592
Spectrometer ${}^2$	531 (476/55)	103	2	0.193974
Image ${}^2$	2000 (1420/580)	140	2	0.07
Fri ${}^2$	1000 (564/436)	101	2	0.101
Scene ${}^3$	2407 (1976/431)	295	2	0.122559
Musk ${}^4$	6598 (5581/1017)	170	2	0.025765
Philippine ${}^1$	5832 (2916/2916)	309	2	0.052984
Ionosphere ${}^4$	351 (126/225)	34	2	0.096866
Optdigits ${}^2$	5620 (572/5048)	64	2	0.011388
Satellite ${}^2$	5100 (75/5025)	37	2	0.007255
Ada ${}^1$	4147 (1029/3118)	49	2	0.011816
Splice ${}^2$	3190 (1535/1655)	62	2	0.019436
HIVA ${}^2$	4229 (149/4080)	1617	2	0.382359

* Dimensionality is the ratio of features to number of observations. Superscripts indicate the data sources as follows: ¹ automl.chalearn.org, ² www.openml.org, ³ mulan.sourceforge.net, ⁴ archive.ics.uci.edu.