% data characteristics
dim = 2; % dimension of the inputs
N = 10000; % number of samples
x = randn(N,dim); % generate data
A = [1 .1; 5 -.1]; % transformation matrix
vN = 0.08; % variance of added noise
nn= randn(N,dim); % noise
% generating the outputs
z = (A * x')' + sqrt(vN)*nn;
% task:
% 1. Generate a THREE dimensional gaussian distribution which contains TWO
% relevant directions.
% You may choose
% the transformation matrix A and
% the additive noise variance
% 2. Display the cloud of sampled points and ensure that MOST of the data
% is in a lower-dimensional subspace.
% 3. Find the relevant directions.