Quasi-regression with shrinkage (Q1873015)

This paper presents a new version of quasi-regression, a method of Monte Carlo approximation useful for global sensitivity analysis, incorporating shrinkage parameters of the type used in wavelet approximation. Section~2 presents the notation for quasi-regression with shrinkage. The tensor product bases studied are described in Section~3. Some elementary properties of the method are recorded in Section~4. Section~5 presents a numerical example in which a black box function from machine learning representing the output of a neural network is approximated. Finally, Section~6 presents conclusions.