Macroscopic Rock Texture Image Classification Using a Hierarchical Neuro-Fuzzy Class Method

Table 8

Performance of neural networks for macro classes of gneiss and basalt; diabase, and rhyolite; basalt, diabase, and rhyolite; gneiss, basalt, diabase, and rhyolite.

Rock	Database test	Correct percentage of training set	Correct percentage of validation set
Gneiss and basalt	Gneiss and basalt1	92.67%	80%
	Gneiss and basalt2	100%	94%
	Gneiss and basalt3	100%	94%
Diلbase and rhyolite	Diabase and rhyolite1	82.4%	72%
	Diabase and rhyolite2	99.2%	85.2%
	Diabase and rhyolite3	95.2%	84.4%
Basalt, diabase, and rhyolite	Basalt, diabase, and rhyolite1	75.14%	60.85%
	Basalt, diabase, and rhyolite2	97.42%	77.42%
	Basalt, diabase, and rhyolite3	99.42%	77.14%
Gneiss, basalt, diabase, and rhyolite	Gneiss, basalt, diabase, and rhyolite1	65.5%	53.75%
	Gneiss, basalt, diabase, and rhyolite2	97%	77%
	Gneiss, basalt, diabase, and rhyolite3	99.75%	68.75%