Covariance Matrix
Is a transformation matrix [see page 5, defined] using the covariance of each pair of dimensions in a sample-space. Transforming with the covariance matrix has the affect of centring the data around the mean.
We define the covariance between two neurons \( kj \) as: \[ C^0_{kj} = \langle (x_k - \langle x_k \rangle)(x_j - \langle x_j \rangle) \rangle \]
Note: The variance is the covariance of an element and itself. \( var(X) = cov(X,X) \)
We can define the covariance-matrix as a 2D matrix \( M[i][j] \) where each cell \( i,j \) contains the covariance between the neurons (\( i,j \)) \( C^0_{ij} \).