ABSTRACT
The coefficients mj must still satisfy equation (22) to preserve the normalization. The noise amplification inherent in applying equation (21) is a defined in equation (26). As before, we want to minimize the quantity
subject to the condition that m·1=1, where 1 is the vector (1, 1, …, 1). As before, w parameterizes the relative weight given to the noise amplification and the fit to the pattern Q(x) .
