Salvato in:
Dettagli Bibliografici
Autori principali: Li, Jiaxing, Chen, Zihan, Chong, Kai Fong Ernest, Das, Bikramjit, Quek, Tony Q. S., Yang, Howard H.
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2409.15100
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916827818360832
author Li, Jiaxing
Chen, Zihan
Chong, Kai Fong Ernest
Das, Bikramjit
Quek, Tony Q. S.
Yang, Howard H.
author_facet Li, Jiaxing
Chen, Zihan
Chong, Kai Fong Ernest
Das, Bikramjit
Quek, Tony Q. S.
Yang, Howard H.
contents Leveraging over-the-air computations for model aggregation is an effective approach to cope with the communication bottleneck in federated edge learning. By exploiting the superposition properties of multi-access channels, this approach facilitates an integrated design of communication and computation, thereby enhancing system privacy while reducing implementation costs. However, the inherent electromagnetic interference in radio channels often exhibits heavy-tailed distributions, giving rise to exceptionally strong noise in globally aggregated gradients that can significantly deteriorate the training performance. To address this issue, we propose a novel gradient clipping method, termed Median Anchored Clipping (MAC), to combat the detrimental effects of heavy-tailed noise. We also derive analytical expressions for the convergence rate of model training with analog over-the-air federated learning under MAC, which quantitatively demonstrates the effect of MAC on training performance. Extensive experimental results show that the proposed MAC algorithm effectively mitigates the impact of heavy-tailed noise, hence substantially enhancing system robustness.
format Preprint
id arxiv_https___arxiv_org_abs_2409_15100
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Robust Federated Learning Over the Air: Combating Heavy-Tailed Noise with Median Anchored Clipping
Li, Jiaxing
Chen, Zihan
Chong, Kai Fong Ernest
Das, Bikramjit
Quek, Tony Q. S.
Yang, Howard H.
Machine Learning
Artificial Intelligence
Leveraging over-the-air computations for model aggregation is an effective approach to cope with the communication bottleneck in federated edge learning. By exploiting the superposition properties of multi-access channels, this approach facilitates an integrated design of communication and computation, thereby enhancing system privacy while reducing implementation costs. However, the inherent electromagnetic interference in radio channels often exhibits heavy-tailed distributions, giving rise to exceptionally strong noise in globally aggregated gradients that can significantly deteriorate the training performance. To address this issue, we propose a novel gradient clipping method, termed Median Anchored Clipping (MAC), to combat the detrimental effects of heavy-tailed noise. We also derive analytical expressions for the convergence rate of model training with analog over-the-air federated learning under MAC, which quantitatively demonstrates the effect of MAC on training performance. Extensive experimental results show that the proposed MAC algorithm effectively mitigates the impact of heavy-tailed noise, hence substantially enhancing system robustness.
title Robust Federated Learning Over the Air: Combating Heavy-Tailed Noise with Median Anchored Clipping
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2409.15100