A new algorithm for the sequential estimation of the regularization parameter in the spline smoothing problem

By expressing the spline smoothing problem in a state-space form, the problem of estimating the regularization parameter is reformulated as a noise variance adaptive estimation problem. The main advantages of the approach proposed with respect to the other state-space methods are a greater simplicity of the algorithm and a reduced computational burden while still retaining the same general underlying hypotheses and the same range of applicability. Numerical results are included.<<ETX>>