A new article about random proximal subspace method with identification-based sampling

Title: Proximal Gradient Methods with Adaptive Subspace Samplings

Together with my supervisors Franck Iutzeler and Jérôme Malick we propose a randomized proximal gradient method that uses enforced by regularizer low dimensional structure of the solution to specify the subspaces and improve the rate in terms of dimensions explored.

Avatar
Dmitry Grishchenko
PhD student in Applied Mathematics