RSSL (version 0.9.3)

MCLinearDiscriminantClassifier: Moment Constrained Semi-supervised Linear Discriminant Analysis.

Description

A linear discriminant classifier that updates the estimates of the means and covariance matrix based on unlabeled examples.

Usage

MCLinearDiscriminantClassifier(X, y, X_u, method = "invariant",
  prior = NULL, x_center = TRUE, scale = FALSE)

Arguments

X

matrix; Design matrix for labeled data

y

factor or integer vector; Label vector

X_u

matrix; Design matrix for unlabeled data

method

character; One of c("invariant","closedform")

prior

Matrix (k by 1); Class prior probabilities. If NULL, estimated from data

x_center

logical; Should the features be centered?

scale

logical; Should the features be normalized? (default: FALSE)

Details

This method uses the parameter updates of the estimated means and covariance proposed in (Loog 2014). Using the method="invariant" option, uses the scale invariant parameter update proposed in (Loog 2014), while method="closedform" using the non-scale invariant version from (Loog 2012).

References

Loog, M., 2012. Semi-supervised linear discriminant analysis using moment constraints. Partially Supervised Learning, LNCS, 7081, pp.32-41.

Loog, M., 2014. Semi-supervised linear discriminant analysis through moment-constraint parameter estimation. Pattern Recognition Letters, 37, pp.24-31.

See Also

Other RSSL classifiers: EMLeastSquaresClassifier, EMLinearDiscriminantClassifier, GRFClassifier, ICLeastSquaresClassifier, ICLinearDiscriminantClassifier, KernelLeastSquaresClassifier, LaplacianKernelLeastSquaresClassifier(), LaplacianSVM, LeastSquaresClassifier, LinearDiscriminantClassifier, LinearSVM, LinearTSVM(), LogisticLossClassifier, LogisticRegression, MCNearestMeanClassifier, MCPLDA, MajorityClassClassifier, NearestMeanClassifier, QuadraticDiscriminantClassifier, S4VM, SVM, SelfLearning, TSVM, USMLeastSquaresClassifier, WellSVM, svmlin()