Variable selection and estimation in the multivariate Functional Linear Model
Angelina Roche

In this talk, we are interested in the problem of variable selection in the multivariate linear functional model. In this model, a quantity of interest depends on several covariates, which can be vectors or elements of a function space (functional data). We propose an estimator minimizing a criterion inspired by the Group-Lasso (Yuan and Lin, 2006) and prove that it verifies a sparsity oracle inequality. An algorithm to approach the solution of the minimization problem without projecting the data will be proposed.