face recognition - Eigenfaces Comlex number Eigenvalues -
when calculate covariance (a_transposed * a) , calculate eigenvalues , eigenvectors 1 small eigenval here have 3 images in training set.
-0.0000000242 17221292.9979712702 11732978.3353619855
this first unsorted eigenval super small. when svd on these eigenvectors need eigenvalues[i]^(-0.5) @ point desired eigenvectors (initial eigenvectors smaller in length since did (a_transposed * a) instead of ( * a_transposed )).
long storyshort, smallest eigenval turns out complex number when math.pow(eval_i,-0.5). suppose use comlex eigensolver ? such small number normal ? other eigenvalues huge, normal?
i have followed this
thanks
Comments
Post a Comment