Towards Active Participant Centric Vertical Federated Learning: Some Representations May Be All You Need

TL;DR

This paper proposes APC-VFL, a novel vertical federated learning approach that efficiently handles partially aligned data among participants with minimal communication, outperforming existing methods in accuracy and cost.

Contribution

Introduces APC-VFL, a new VFL method that requires only one communication step and effectively manages unaligned data through local representation learning and knowledge distillation.

Findings

APC-VFL outperforms SplitNN and VFedTrans in accuracy and communication costs.

APC-VFL maintains high performance with decreasing data alignment ratios.

The method simplifies VFL operations with a single communication step.

Abstract

Existing Vertical FL (VFL) methods often struggle with realistic and unaligned data partitions, and incur into high communication costs and significant operational complexity. This work introduces a novel approach to VFL, Active Participant Centric VFL (APC-VFL), that excels in scenarios when data samples among participants are partially aligned at training. Among its strengths, APC-VFL only requires a single communication step with the active participant. This is made possible through a local and unsupervised representation learning stage at each participant followed by a knowledge distillation step in the active participant. Compared to other VFL methods such as SplitNN or VFedTrans, APC-VFL consistently outperforms them across three popular VFL datasets in terms of F1, accuracy and communication costs as the ratio of aligned data is reduced.