Concentration estimates for the moving least-square method in learning theory (Q719353): Difference between revisions

Moving least-square method is a classical tool in statistics, numerical analysis and approximation theory. It was studied by the author, \textit{D.-H. Xiang} and \textit{D.-X. Zhou} [ J. Approx. Theory 162, No. 3, 599--614 (2010; Zbl 1202.68325)] in a learning theory framework. In this paper the author improves the estimate for the sampling error by means of a concentration inequality, which leads to better convergence rates.