Orateur
Léo Gonin
Description
We propose a kernel-based estimation framework for multivariate generalized gamma convolutions (MGGC), a class of probability distributions on Rd +. MGGC are useful, for example, in risk modeling, but their densities are generally not explicit. The construction relies on a new Stein operator for the MGGC class, derived from a partial differential equation satisfied by the characteristic function and expressed as an integral against the underlying Thorin measure. This leads to the construction of a new kernel and an estimation procedure in the resulting reproducing kernel Hilbert space (RKHS), together with consistency results. Numerical experiments illustrate the practical behaviour of the proposed estimator.