Single-Round Scalable Analytic Federated Learning

TL;DR

SAFLe introduces a scalable, non-linear federated learning framework that maintains single-round communication, achieves high accuracy, and is mathematically equivalent to linear regression, outperforming existing methods.

Contribution

SAFLe breaks the linear-nonlinear trade-off in analytic federated learning by enabling scalable non-linear models within a single-round framework.

Findings

SAFLe outperforms linear AFL and multi-round DeepAFL in accuracy.

SAFLe is mathematically equivalent to high-dimensional linear regression.

SAFLe achieves state-of-the-art results in federated vision benchmarks.

Abstract

Federated Learning (FL) is plagued by two key challenges: high communication overhead and performance collapse on heterogeneous (non-IID) data. Analytic FL (AFL) provides a single-round, data distribution invariant solution, but is limited to linear models. Subsequent non-linear approaches, like DeepAFL, regain accuracy but sacrifice the single-round benefit. In this work, we break this trade-off. We propose SAFLe, a framework that achieves scalable non-linear expressivity by introducing a structured head of bucketed features and sparse, grouped embeddings. We prove this non-linear architecture is mathematically equivalent to a high-dimensional linear regression. This key equivalence allows SAFLe to be solved with AFL's single-shot, invariant aggregation law. Empirically, SAFLe establishes a new state-of-the-art for analytic FL, significantly outperforming both linear AFL and multi-round DeepAFL in accuracy across all benchmarks, demonstrating a highly efficient and scalable solution for federated vision.