Graph-Aware Learning Rates for Decentralized Optimization

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Fainman, Aaron, Vlaski, Stefan
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918339095298048
author Fainman, Aaron
Vlaski, Stefan
author_facet Fainman, Aaron
Vlaski, Stefan
contents We propose an adaptive step-size rule for decentralized optimization. Choosing a step-size that balances convergence and stability is challenging. This is amplified in the decentralized setting as agents observe only local (possibly stochastic) gradients and global information (like smoothness) is unavailable. We derive a step-size rule from first principles. The resulting formulation reduces to the well-known Polyak's rule in the single-agent setting, and is suitable for use with stochastic gradients. The method is parameter free, apart from requiring the optimal objective value, which is readily available in many applications. Numerical simulations demonstrate that the performance is comparable to the optimally fine-tuned step-size.
format Preprint
id arxiv_https___arxiv_org_abs_2509_14854
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Graph-Aware Learning Rates for Decentralized Optimization
Fainman, Aaron
Vlaski, Stefan
Optimization and Control
Signal Processing
We propose an adaptive step-size rule for decentralized optimization. Choosing a step-size that balances convergence and stability is challenging. This is amplified in the decentralized setting as agents observe only local (possibly stochastic) gradients and global information (like smoothness) is unavailable. We derive a step-size rule from first principles. The resulting formulation reduces to the well-known Polyak's rule in the single-agent setting, and is suitable for use with stochastic gradients. The method is parameter free, apart from requiring the optimal objective value, which is readily available in many applications. Numerical simulations demonstrate that the performance is comparable to the optimally fine-tuned step-size.
title Graph-Aware Learning Rates for Decentralized Optimization
topic Optimization and Control
Signal Processing
url https://arxiv.org/abs/2509.14854