Description demo_unsupervised_ISOMAP.m

Demonstrates use of ISOMAP to visualize a dataset in lower dimensions

clear all
close all
load animals.mat
[n,d] = size(X);

usage of KPCA with rbf basis

Reduce to 2-dimensions with KPCA

kernelArgs = struct('sigma',10);
options = struct('maxComponents',2,'kernelFunc',@ml_kernel_rbf,...
model = ml_unsupervised_dimRedKPCA(X,options);
Xreduced = model.reduceFunc(model,X);
grid on
grid minor
title('KPCA Projection onto 2-dimensions of animals data (rbf kernel)');
Number of Components selected: 2
Variance explained by basis: 0.22

usage ISOMAP to visualize animals dataset in low dimensions

options = [];
options.K = 2;
options.names = animals;
options.disconnected = 1;