Properties Also known as Karhunen-Loeve transform (KLT) Works for distances only as others are not invariant to rotation Mean-centering   Easier way to compute covariance: is covariance matrix   Allows use of SVD to compute PCA Can be done using SVD   Eigenvector matrix of is really the SVD transform matrix for   Different from SVD of though How many dimensions to retain?   Based on energy (similar to SVD)   Total energy is sum of eigenvalues   Retain dimensions such that of the energy is retained   In the above example, = 1 retains of the energy Running time: for of size