Support vector machines

Support vector machines (SVMs) (52) seek cuts of the data that separate classes effectively, that is by large gaps. Technically, SVMs operate by finding a hypersurface in the space of gene expression profiles, that will split the groups so that there is the largest distance between the hypersur-face and the nearest of the points in the groups. More flexible implementations allow for imperfect separation of groups. See Burges (53) and Christianini and Shawe-Taylor (54) for details of SVMs and generalizations, while Lee and Lee (55) and Brown et al. (56) give examples of analysis of gene expression data using SVMs.

0 0

Post a comment