QUADRC

Trainable quadratic classifier

    W = QUADRC(A,R,S)
    W = A*QUADRC([],R,S)
    W = A*QUADRC(R,S)

Input
A	Dataset
R,S	0 <= R,S <= 1, regularisation parameters (default: R = 0, S = 0)

Output
W	Quadratic Discriminant Classifier mapping

Description

Computation of the quadratic classifier between the classes of the dataset  A based on different class covariances. R and S are regularisation  parameters used for finding the covariance matrix as

      G = (1-R-S)*G + R*diag(diag(G)) + S*mean(diag(G))*eye(size(G,1))

This routine differs from QDC; instead of using the densities, it is based  on the class covariances and does not use class priors. The multi-class  problem is solved by multiple two-class quadratic discriminants, using  MCLASSC.

See also

mappings, datasets, fisherc, nmc, nmsc, ldc, udc, qdc, mclassc, fishercc,