A maximum likelihood (ML) method which determines the regularization parameter optimally was constructed the construction of the method is a simple extension of the ideas of Anderssen and Bloomfield, who consider the problem of numerically differentiating noisy data.