Stepwise iterative maximum likelihood clustering approach

A clustering algorithm using iterative maximum likelihood

% [Cluster,MaxLtot,DelLtot] = SIML(X,class,var,MaxIteration,InitMethod,Repeat)
%
% [Cluster,MaxLtot,DelLtot] = SIML(X) or SOML(X,class) or SOML(X,class,var)
% or SIML(X,class,var,MaxIteration)
% or SIML(X,class,var,MaxIteration,InitMethod)
%
% Stepwsie optimal maximum likelihood method
%
% INPUT
% 1) X <- number of samples x dimension (Data)
%
% 2) class <- number of classes
% or a range of class e.g. class = 1:5
% If no class information is given then default value 1:5 will be used.
%
% 3) var: default 0 (no figures)
% 1 (all plots - time consuming)
% 2 (only final cluster plot, MaxLtot plot and DelLtot plot)
%
% 4) MaxIteration <- max iteration before algorithm is exited (default 15)
%
% 5) InitMethod <- 1 for Random Initialization
% 2 for kmeans Initialization (default is 2)
% 3 iterative (supply means of c-1 classes for c-class)
% 4 for Max Min of norm of data
%
% 6) Repeat <- number of time the algorithm is repeated to find the best
% solution. Default value is 10. Increasing the value of
% Repeat may increase the clustering accuracy.
%
% OUTPUT
% 1) Cluster is labels of samples (number of samples x class range)
% 2) MaxLtot plot
% 3) DelLtot plot
%
% NOTE: Only meant for small dimension; i.e., d < n
%
% Alok Sharma, RIKEN, Japan; 3-Jun-2015
% Ref: Sharma et al., Stepwise iterative maximum likelihood clustering approach, BMC Bioinformatics, 17(319), 1-14, 2016