Multi-Tier Federated Learning for Vertically Partitioned Data

TL;DR

This paper introduces TDCD, a communication-efficient decentralized algorithm for tiered federated learning on vertically partitioned data, with theoretical convergence analysis and empirical validation.

Contribution

It proposes a novel tiered decentralized coordinate descent algorithm tailored for vertically partitioned data in multi-tier networks, with convergence guarantees.

Findings

Convergence rate depends on vertical partitions, local updates, and clients.

Algorithm reduces communication overhead through local gradient steps.

Empirical results validate effectiveness on various datasets and objectives.

Abstract

We consider decentralized model training in tiered communication networks. Our network model consists of a set of silos, each holding a vertical partition of the data. Each silo contains a hub and a set of clients, with the silo's vertical data shard partitioned horizontally across its clients. We propose Tiered Decentralized Coordinate Descent (TDCD), a communication-efficient decentralized training algorithm for such two-tiered networks. To reduce communication overhead, the clients in each silo perform multiple local gradient steps before sharing updates with their hub. Each hub adjusts its coordinates by averaging its workers' updates, and then hubs exchange intermediate updates with one another. We present a theoretical analysis of our algorithm and show the dependence of the convergence rate on the number of vertical partitions, the number of local updates, and the number of clients in each hub. We further validate our approach empirically via simulation-based experiments using a variety of datasets and both convex and non-convex objectives.