I am trying to center a kernel matrix for PCA. It says I can center it using the equation
K = (K − 1N*K − K*1N + 1N*K*1N) , 1N ∈ RN×N, (1N)ij := 1
N Is the new K also a covariance matrix? Normally I would subtract the mean of rows from X, and then find the covariance matrix. here the new K is directly used to find eigenvectors. So does this mean K is the covariance matrix too