Parameter-Efficient and Personalized Federated Training of Generative Models at the Edge

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Khan, Kabir, Sarkar, Manju, Kar, Anita, Ghosh, Suresh
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917173112340480
author Khan, Kabir
Sarkar, Manju
Kar, Anita
Ghosh, Suresh
author_facet Khan, Kabir
Sarkar, Manju
Kar, Anita
Ghosh, Suresh
contents Large generative models (for example, language and diffusion models) enable high-quality text and image synthesis but are hard to train or adapt in cross-device federated settings due to heavy computation and communication and statistical/system heterogeneity. We propose FedGen-Edge, a framework that decouples a frozen, pre-trained global backbone from lightweight client-side adapters and federates only the adapters. Using Low-Rank Adaptation (LoRA) constrains client updates to a compact subspace, which reduces uplink traffic by more than 99 percent versus full-model FedAvg, stabilizes aggregation under non-IID data, and naturally supports personalization because each client can keep a locally tuned adapter. On language modeling (PTB) and image generation (CIFAR-10), FedGen-Edge achieves lower perplexity/FID and faster convergence than strong baselines while retaining a simple FedAvg-style server. A brief ablation shows diminishing returns beyond moderate LoRA rank and a trade-off between local epochs and client drift. FedGen-Edge offers a practical path toward privacy-preserving, resource-aware, and personalized generative AI on heterogeneous edge devices.
format Preprint
id arxiv_https___arxiv_org_abs_2511_11585
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Parameter-Efficient and Personalized Federated Training of Generative Models at the Edge
Khan, Kabir
Sarkar, Manju
Kar, Anita
Ghosh, Suresh
Machine Learning
Distributed, Parallel, and Cluster Computing
I.2.6; I.2.7; C.2.4
Large generative models (for example, language and diffusion models) enable high-quality text and image synthesis but are hard to train or adapt in cross-device federated settings due to heavy computation and communication and statistical/system heterogeneity. We propose FedGen-Edge, a framework that decouples a frozen, pre-trained global backbone from lightweight client-side adapters and federates only the adapters. Using Low-Rank Adaptation (LoRA) constrains client updates to a compact subspace, which reduces uplink traffic by more than 99 percent versus full-model FedAvg, stabilizes aggregation under non-IID data, and naturally supports personalization because each client can keep a locally tuned adapter. On language modeling (PTB) and image generation (CIFAR-10), FedGen-Edge achieves lower perplexity/FID and faster convergence than strong baselines while retaining a simple FedAvg-style server. A brief ablation shows diminishing returns beyond moderate LoRA rank and a trade-off between local epochs and client drift. FedGen-Edge offers a practical path toward privacy-preserving, resource-aware, and personalized generative AI on heterogeneous edge devices.
title Parameter-Efficient and Personalized Federated Training of Generative Models at the Edge
topic Machine Learning
Distributed, Parallel, and Cluster Computing
I.2.6; I.2.7; C.2.4
url https://arxiv.org/abs/2511.11585