Preprocessing for ICA

The preprocessing of ICA can be done as follows:

• Centering: The first step is to center x. That is, we need to subtract its mean vector from x so as to make x a zero-mean variable.
• Whitening: Before putting the data through ICA, we are supposed to whiten the data. This means that the data has to be uncorrelated. Geometrically speaking, it tends to restore the initial shape of the data and only the resultant matrix needs to be rotated.