Journal of Big Data

Table 8 Dataset V, and test 3 parameters

From: A graphical heuristic for reduction and partitioning of large datasets for scalable supervised training

Parameter of	Parameter	Test 3
Dataset	# of data points	6668
	Feature extraction	VGG16
	Training/total ratio	0.75
	# features	25 088
Classification algorithm		LIBSVM
Step 1	\(n_{c}\)	166
Step 2	\(C_I\)	\(e^{1.0}\)
Step 2	\(C_E\)	\(e^{4.0}\)
Step 3	GS edge cut	7.01
LIBSVM	Training time (sec)	328.39
GSH/Speed-up	Training time (sec)	70.51/4.7
LIBSVM	Prediction accuracy (%)	99.16
GSH	Prediction accuracy (%)	99.16

Back to article page