Faster to the Wrong Answer: The Price of Local Updates in Federated Optimization
Organizations: Imperial Business School
Abstract
Local computation can improve the conditioning of federated updates but may also introduce bias, causing their fixed point to differ from the global minimizer. We characterize the minimum worst-case bias achievable under a prescribed conditioning bound. We consider strongly convex quadratic client objectives with Hessian eigenvalues in and local updates represented by a shared positive spectral filter. Client heterogeneity is measured by the root-mean-square gradient at the global minimizer. For a uniform surrogate condition number bounded by , we establish the exact worst-case bias per unit heterogeneity, , and show that local gradient descent with sufficiently small step sizes and proximal updates attain this bound. Our analysis reduces populations of arbitrary dimension, including noncommuting Hessians, to distributions over scalar curvatures. This reduction provides optimality certificates and worst-case constructions involving at most three scalar clients. We further extend the sharp proximal bias bound to smooth strongly convex objectives beyond quadratics and quantify the error introduced by inexact local solves.
Figures & tables
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
| Quantity | Recorded value |
|---|---|
| Scalar witness configurations | 74 |
| Maximum witness discrepancy | |
| Noncommuting matrix populations | 400 |
| Maximum matrix bias / sharp uniform bound | 0.904697 |
| Maximum matrix identity discrepancy | |
| Nonquadratic residual checks | 100 |
Federated Learning
Federated learning trains shared models across data holders without requiring them to pool their raw data.
See all 1,037 papers