Table 3.

Performance comparison of the three-feature SVM classifier to linear classifiers, an RBF network classifier and other SVM classifiers, using canonical training and testing datasets.

		Accuracy
	Feature(s)	Canonical testing dataset	Homologous-regions-only testing dataset
3-feature SVM classifier	Sequence similarity, inverse CBIN count, match/mismatch fraction (cf. Table 2)	99.63%	98.98%
2-feature SVM classifiers	Match/mismatch fraction, sequence similarity	97.50%	96.68%
	Inverse CBIN count, sequence similarity	99.32%	98.97%
	Match/mismatch fraction, inverse CBIN count	99.42%	98.91%
RBF Network classifier	Sequence similarity, inverse CBIN count, match/mismatch fraction	99.32%	98.79%
3-feature linear classifier	Sequence similarity, inverse CBIN count, match/mismatch fraction	99.42%	98.80%
2-feature linear classifiers	Match/mismatch fraction, sequence similarity	99.03%	98.75%
	Inverse CBIN count, sequence similarity	99.32%	98.67%
	Match/mismatch fraction, inverse CBIN count	99.37%	98.77%
1-feature linear classifiers	Sequence similarity	82.22%	82.02%
	Match/mismatch fraction	98.05%	98.62%
	Inverse CBIN count	99.37%	98.75%