ORKMeans {ORKM}

Online regularized K-means clustering algorithm for online multi-view data

Description

For the online clustering problem, this function proposes the Online Regularized K-means Clustering (ORKMC) method to deal with online multi-view data. Firstly, for the clustering problem of multi-view data, a non-negative matrix decomposition is used as the starting point of the model to find the indicator matrix and cluster centres of each cluster; for online updating, a projected gradient descent method is proposed to perform online updating to improve the accuracy and speed of data clustering; for the overfitting phenomenon, regularisation is proposed to avoid the above problem. In addition, since the choice of regularization parameters is extremely important to the effectiveness of the ORKMC algorithm, the choice of regularization parameters varies in different datasets. In this paper, a suitable range of regularisation parameters and model parameters is given. The effectiveness of the ORKMC algorithm is tested through an extensive study of multi-view/single-view data. The validity of the ORKMC algorithm is tested through an extensive study of multi-view/single-view data.

Usage

ORKMeans(X,K,V,chushi,r,yita,gamma,alpha,epsilon,truere,max.iter,method=0)

Arguments

X

is the online single/multi-view data matrix

K

is the number of cluster

V

is the view of X

chushi

is the initial value for online

yita

is the regularized parameter

r

is the banlance parameter

gamma

is the step size

alpha

is the caculated the weight of view

epsilon

is the epsilon

truere

is the ture label in data set

max.iter

is the max iter

method

is the caluate the NMI

Value

NMI,weight,center,result

Author(s)

Miao Yu

Examples

library(MASS) 
library(Matrix)  
  yita=0.5;V=2;chushi=100;K=3;r=0.5;max.iter=10;n1=n2=n3=70;gamma=0.1;alpha=0.98;epsilon=1
  X1<-rnorm(n1,20,2);X2<-rnorm(n2,25,1.5);X3<-rnorm(n3,30,2) 
  Xv<-c(X1,X2,X3)
  data<-matrix(Xv,n1+n2+n3,2)
  data[1:70,2]<-1;data[71:140,2]<-2;data[141:210,2]<-3
  truere=data[,2]
  X<-matrix(data[,1],n1+n2+n3,1) 
  lamda1<-0.2;lamda2<-0.8
  lamda<-matrix(c(lamda1,lamda2),nrow=1,ncol=2)
  sol.svd <- svd(lamda)
  U1<-sol.svd$u
  D1<-sol.svd$d
  V1<-sol.svd$v
  C1<-t(U1)
  Y1<-C1/D1
  view<-V1
  view1<-matrix(view[1,])
  view2<-matrix(view[2,])
  X1<-matrix(view1,n1+n2+n3,1)
  X2<-matrix(view2,n1+n2+n3,1)
  ORKMeans(X=X1,K=K,V=V,r=r,chushi=chushi,yita=yita,gamma=gamma,epsilon=epsilon,
max.iter=max.iter,truere=truere,method=0)