Cross-Silo Federated Learning for Multi-Tier Networks with Vertical and Horizontal Data Partitioning

Abstract

We consider federated learning 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. The clients in each silo perform multiple local gradient steps before sharing updates with their hub to reduce communication overhead. 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 and the number of local updates. We further validate our approach empirically via simulation-based experiments using a variety of datasets and objectives.

How Green Credit Policy Affects Commercial Banks' Credit Risk?

“…This paper presents a federated learning model that utilizes seven indicators, such as the balance of green credit business and transaction information data from commercial banks. Financial institutions, such as banks and consumer finance companies, integrate their own accumulated corporate green financial behavior information with external data related to the company to establish an accurate model (Wu et al, 2020;Das et al, 2022). The risk control model aims to solve problems such as white accounts in credit investigation and green enterprise loan difficulties caused by a lack of effective data, thereby effectively improving the credit risk control level of banks (Sattler et al, 2021;Zhang et al, 2021).…”