1: \begin{abstract}

2: We consider a nonparametric additive model of a conditional mean

3: function in which the number of variables and additive components may

4: be larger than the sample size but the number of nonzero additive

5: components is ``small'' relative to the sample size. The statistical

6: problem is to determine which additive components are nonzero. The

7: additive components are approximated by truncated series expansions

8: with B-spline bases. With this approximation, the problem of component

9: selection becomes that of selecting the groups of coefficients in the

10: expansion. We apply the adaptive group Lasso to select nonzero

11: components, using the group Lasso to obtain an initial estimator and

12: reduce the dimension of the problem. We give conditions under which

13: the group Lasso selects a model whose number of components is

14: comparable with the underlying model, and the adaptive group Lasso

15: selects the nonzero components correctly with probability approaching

16: one as the sample size increases and achieves the optimal rate of

17: convergence. The results of Monte Carlo experiments show that the

18: adaptive group Lasso procedure works well with samples of moderate

19: size. A data example is used to illustrate the application of the

20: proposed method.

21: \end{abstract}