Combining Dissimilarities in a Hyper Reproducing Kernel Hilbert Space for Complex Human Cancer Prediction

Table 3

Accuracy for the 𝜈-SVM using a linear combination of non-Euclidean dissimilarities in an HRKHS. The 𝜈-SVM based on the best distance, the classical -SVM, and the Lanckriet formalism have been taken as a reference.

Technique	Breast B	DLBCL C	DLBCL D
𝜈-SVM (Coordinates)	10.20%	6.89%	12.96%
𝜈-SVM (Best Distance)	8.6%	6.89%	14.81%
𝜈-SVM (Nonlinear kernel)	8.16%	6.89%	12.96%
Lanckriet (finite family)	8%	10.3%	25.2%
Infinite family of distances	6%	5.33%	16%