Table 1.

Datasets characteristics

Experiments type	Dataset		Dataset properties
			Domain	Data type	Features [#]	Train samples [#]	Test samples[#]
RBMs variants	UCI evaluation suite65	ADULT	Households	Binary	123	5000	26,147
		Connect4	Games	Binary	126	16,000	47,557
		DNA	Genetics	Binary	180	1400	1186
		Mushrooms	Biology	Binary	112	2000	5624
		NIPS-0-12	Documents	Binary	500	400	1240
		OCR-letters	Letters	Binary	128	32,152	10,000
		RCV1	Documents	Binary	150	40,000	150,000
		Web	Internet	Binary	300	14,000	32,561
	CalTech 101 Silhouettes66	16 × 16	Images	Binary	256	4082	2302
		28 × 28	Images	Binary	784	4100	2307
	MNIST67		Digits	Binary	784	60,000	10,000
MLPs variants	MNIST67		Digits	Grayscale	784	60,000	10,000
	CIFAR1068		Images	RGB colors	3072	50,000	10,000
	HIGGS1		Particle physics	Real values	28	10,500,000	500,000
	Fashion-MNIST69		Fashion products	Grayscale	784	60,000	10,000
CNNs variants	CIFAR1068		Images	RGB colors	3072	50,000	10,000

The data used in this paper have been chosen to cover a wide range of fields where ANNs have the potential to advance state-of-the-art, including biology, physics, computer vision, data mining, and economics