Correlation Kernels for Support Vector Machines Classification with Applications in Cancer Data

Table 3

Prediction accuracy comparison among different algorithms for Leukemia cancer.

DT	UC			PC	DCB	KNN
	𝛾=0.1	𝛾=1	𝛾=10			𝐾=1	𝐾=5	𝐾=10
0.8189	0.7111	0.7868	0.7111	0.9150	𝟎.𝟗𝟐𝟓𝟕	0.8443	0.8557	0.8082

Here DT means decision tree algorithm, UC means the usual correlation kernel, PC stands for the parsimonious correlation we proposed, DCB is the novel kernel proposed by us representing denoised correlation-based kernel, and KNN is the K-nearest neighborhood algorithm.