RSSL (version 0.9.3)

KernelICLeastSquaresClassifier: Kernelized Implicitly Constrained Least Squares Classification

Description

A kernel version of the implicitly constrained least squares classifier, see ICLeastSquaresClassifier.

Usage

KernelICLeastSquaresClassifier(X, y, X_u, lambda = 0,
  kernel = vanilladot(), x_center = TRUE, scale = TRUE, y_scale = TRUE,
  lambda_prior = 0, classprior = 0, method = "LBFGS",
  projection = "semisupervised")

Arguments

X

matrix; Design matrix for labeled data

y

factor or integer vector; Label vector

X_u

matrix; Design matrix for unlabeled data

lambda

numeric; L2 regularization parameter

kernel

kernlab::kernel to use

x_center

logical; Should the features be centered?

scale

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

y_scale

logical; whether the target vector should be centered

lambda_prior

numeric; regularization parameter for the posterior deviation from the prior

classprior

The classprior used to compare the estimated responsilibities to

method

character; Estimation method. One of c("LBFGS")

projection

character; The projection used. One of c("supervised","semisupervised")