1: \begin{abstract}

2: We propose stochastic algorithms for solving large scale nonsmooth

3: convex composite minimization problems. They activate at

4: each iteration blocks of randomly selected proximity operators and

5: achieve almost sure convergence of the iterates to a solution

6: without any regularity assumptions. Numerical applications to data

7: analysis problems are provided.

8: \end{abstract}

9: