pca.m

来自「A toolbox for the non-local means algori」· M 代码 · 共 26 行

function [Y,X1,v,Psi] = pca(X,numvecs, options)

% pca - compute the principal component analysis.
%
%   [Y,X1,v,Psi] = pca(X,numvecs)
%
%   X is a matrix of size dim x p of data points.
%   X1 is the matrix of size numvecs x p (projection on the numvect first eigenvectors)
%	Y the matrix  of size dim x numvecs of numvecs first eigenvector of the correlation matrix X*X'
%		(this matrix is computed using the traditional flipping trick if p is large).
%	v is the vector of size numvecs of eigenvalues.
%   Psi is the mean.
%
%   Warning: the mean of X is substracted before computing the covariance
%   matrix.
%
%   You can use an iterative algorithm based 
%   on expectation maximization by setting
%       options.use_em = 1;
%   if you want a fast estimation of a few eigenvectors.
%   This algorithm use the code of Sam Roweis
%       Sam Roweis, "EM Algorithms for PCA and SPCA",
%       Neural Information Processing Systems 10 (NIPS'97) pp.626-632
%       http://www.cs.toronto.edu/~roweis/code.html
%
%   Copyright (c) 2006 Gabriel Peyr