The Mirror-Prox Sliding Method for Non-smooth decentralized saddle-point problems
Ilya Kuruzov, Alexander Rogozin, Demyan Yarmoshik, Alexander Gasnikov
TL;DR
This work obtains an algorithm for non-smooth decentralized saddle-point problems through a specific penalization method and this sliding, which approaches lower bounds for several communication rounds and calls of (sub-)gradient per node.
Abstract
The saddle-point optimization problems have a lot of practical applications. This paper focuses on such non-smooth problems in decentralized case. This work contains generalization of recently proposed sliding for centralized problem. Through specific penalization method and this sliding we obtain algorithm for non-smooth decentralized saddle-point problems. Note, the proposed method approaches lower bounds both for number of communication rounds and calls of (sub-)gradient per node.