Efficient Determination of Optimal Smoothing Parameters in Spline Approximations

摘要

In this paper, we present a systematic method for determining an optimal smoothing parameter, an important issue in the design of smoothing splines. The optimal parameter is obtained as the minimizer of the so-called cross-validation function, and the global minimum is found very efficiently using extrema detection method of spline functions. Compared to the conventional search algorithm, the computational amount is reduced dramatically. The effectiveness is verified by a numerical example.