Compressed-VFL: Communication-Efficient Learning with Vertically Partitioned Data
This work addresses communication efficiency for distributed machine learning with vertically partitioned data, representing an incremental improvement over existing VFL methods.
The paper tackles the problem of high communication costs in vertical federated learning by proposing Compressed-VFL, which uses compression techniques to reduce communication by over 90% without significantly harming accuracy.
We propose Compressed Vertical Federated Learning (C-VFL) for communication-efficient training on vertically partitioned data. In C-VFL, a server and multiple parties collaboratively train a model on their respective features utilizing several local iterations and sharing compressed intermediate results periodically. Our work provides the first theoretical analysis of the effect message compression has on distributed training over vertically partitioned data. We prove convergence of non-convex objectives at a rate of $O(\frac{1}{\sqrt{T}})$ when the compression error is bounded over the course of training. We provide specific requirements for convergence with common compression techniques, such as quantization and top-$k$ sparsification. Finally, we experimentally show compression can reduce communication by over $90\%$ without a significant decrease in accuracy over VFL without compression.