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

[摘要] DNA microarrays provide rich profiles that are used incancer prediction considering the gene expression levelsacross a collection of related samples. Support Vector Machines(SVM) have been applied to the classification of cancersamples with encouraging results. However, they rely onEuclidean distances that fail to reflect accurately the proximitiesamong sample profiles. Then, non-Euclidean dissimilaritiesprovide additional information that should be consideredto reduce the misclassification errors. In this paper, we incorporate in theν-SVM algorithm alinear combination of non-Euclidean dissimilarities. Theweights of the combination are learnt in a (HyperReproducing Kernel Hilbert Space) HRKHS using a SemidefiniteProgramming algorithm. This approach allows us to incorporatea smoothing term that penalizes the complexity of thefamily of distances and avoids overfitting. The experimental results suggest that the method proposedhelps to reduce the misclassification errors in severalhuman cancer problems.

[学科分类] 基础医学