Sampling from non-smooth distribution through Langevin diffusion

Duy Tung Luu (1) , Jalal M. Fadili (1) , Christophe Chesneau (2)

1 Equipe Image - Laboratoire GREYC - UMR6072
2 LMNO - Laboratoire de Mathématiques Nicolas Oresme

In this paper, we propose algorithms for sampling from the distributions whose density is non-smoothed nor log-concave. Our algorithms are based on the Langevin diffusion on the regularized counterpart of density by the Moreau-Yosida regularization. These results are then applied for computing the exponentially weighted aggregates for high dimensional regression.

Mots clés: high dimensional regression; Moreau Langevin diffusion; Yosida smoothing; exponential weighted aggregation (EWA); sparse learning

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]

Duy Tung Luu, Jalal M. Fadili, Christophe Chesneau. Sampling from non-smooth distribution through Langevin diffusion. ORASIS 2017, GREYC, Jun 2017, Colleville-sur-Mer, France. ⟨hal-01866621⟩