Sample-based Federated Learning via Mini-batch SSCA
This work addresses federated learning challenges for distributed systems, but it appears incremental as it applies an existing technique (SSCA) to a new context.
The paper tackles federated optimization with nonconvex constraints by proposing privacy-preserving algorithms based on stochastic successive convex approximation (SSCA), showing convergence to KKT points and demonstrating advantages in convergence speed, communication cost, and model specification through numerical experiments.
In this paper, we investigate unconstrained and constrained sample-based federated optimization, respectively. For each problem, we propose a privacy preserving algorithm using stochastic successive convex approximation (SSCA) techniques, and show that it can converge to a Karush-Kuhn-Tucker (KKT) point. To the best of our knowledge, SSCA has not been used for solving federated optimization, and federated optimization with nonconvex constraints has not been investigated. Next, we customize the two proposed SSCA-based algorithms to two application examples, and provide closed-form solutions for the respective approximate convex problems at each iteration of SSCA. Finally, numerical experiments demonstrate inherent advantages of the proposed algorithms in terms of convergence speed, communication cost and model specification.