DIMIX: DIminishing MIXing for Sloppy Agents

Abstract

We study non-convex distributed optimization problems where a set of agents collaboratively solve a separable optimization problem that is distributed over a time-varying network. The existing methods to solve these problems rely on (at most) one time-scale algorithms, where each agent performs a diminishing or constant step-size gradient descent at the average estimate of the agents in the network. However, if possible at all, exchanging exact information, that is required to evaluate these average estimates, potentially introduces a massive communication overhead. Therefore, a reasonable practical assumption to be made is that agents only receive a rough approximation of the neighboring agents' information. To address this, we introduce and study a two time-scale decentralized algorithm with a broad class of lossy information sharing methods (that includes noisy, quantized, and/or compressed information sharing) over time-varying networks. In our method, one time-scale suppresses the (imperfect) incoming information from the neighboring agents, and one time-scale operates on local cost functions' gradients. We show that with a proper choices for the step-sizes' parameters, the algorithm achieves a convergence rate of O(T −1/3+ ) for non-convex distributed optimization problems over time-varying networks, for any > 0. Our simulation results support the theoretical results of the paper.

Almost Sure Convergence of Distributed Optimization with Imperfect Information Sharing

“…In this paper, we study the distributed convex optimization problems over time-varying networks with imperfect information sharing. We consider the two-time-scale gradient descent method studied in [18,22] to solve the optimization problem. One time-scale adjusts the (imperfect) incoming information from the neighboring agents, and one time-scale controls the local cost functions' gradients.…”