mkkcEst: An internal function called 'mkkcEst'.

In SeojinBang/MKKC: Multiple Kernel K-means Clustering

Description

It performs multiple kernel K-means clustering on a multi-view data.

Usage

mkkcEst(K, centers, iter.max = 10, A = NULL, bc = NULL,
  epsilon = 1e-04, theta = rep(1/dim(K)[3], dim(K)[3]))

Arguments

K	N x N x P array containing P kernel matrices with size N x N.
centers	The number of clusters, say k.
iter.max	The maximum number of iterations allowed. The default is 10.
A	Linear constraint matrix.
bc	Lower and upper constraint bounds.
epsilon	Convergence threshold. The default is 10^{-4}.
theta	intial values for kernel coefficients. The default is 1/P for all views.

Value

mkkcEst returns the following components:

cluster

A vector of integers (from 1:k) indicating the cluster to which each point is allocated.

totss

The total sum of squares.

withinss

Matrix of within-cluster sum of squares by cluster, one row per view.

withinsscluster

Vector of within-cluster sum of squares, one component per cluster.

withinssview

Vector of within-cluster sum of squares, one component per view.

tot.withinss

Total within-cluster sum of squares, i.e. sum(withinsscluster).

betweenssview

Vector of between-cluster sum of squares, one component per view.

tot.betweenss

The between-cluster sum of squares, i.e. totss-tot.withinss.

clustercount

The number of clusters, say k.

coefficients

The kernel coefficients

H

The continuous clustering assignment

size

The number of points, one component per cluster.

iter

The number of iterations.

SeojinBang/MKKC documentation built on Sept. 18, 2019, 1:42 p.m.