Covariance Matrices

• For each pair of input dimensions we can compute the covariance of their values over a set of data vectors. • These covariances can be arranged in a symmetric matrix. • The terms along the diagonal are the variances of the individual input dimensions.

i      j     k

i j k