Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yang, Xiaohong, Xie, Tong, Liwang, Minghui, Shang, Chikai, Lu, Yang, Jiao, Zhenzhen, Fu, Liqun, Hosseinalipour, Seyyedali
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2601.13824
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918377393487872
author Yang, Xiaohong
Xie, Tong
Liwang, Minghui
Shang, Chikai
Lu, Yang
Jiao, Zhenzhen
Fu, Liqun
Hosseinalipour, Seyyedali
author_facet Yang, Xiaohong
Xie, Tong
Liwang, Minghui
Shang, Chikai
Lu, Yang
Jiao, Zhenzhen
Fu, Liqun
Hosseinalipour, Seyyedali
contents Training large language models (LLMs) at the network edge faces fundamental challenges arising from device resource constraints, severe data heterogeneity, and heightened privacy risks. To address these challenges, we propose ELSA (Efficient LLM-centric Split Aggregation), a novel framework that systematically integrates split learning (SL) and hierarchical federated learning (HFL) for distributed LLM fine-tuning over resource-constrained edge networks. ELSA introduces three key innovations. First, it employs a task-agnostic, behavior-aware client clustering mechanism that constructs semantic fingerprints using public probe inputs and symmetric Kullback-Leibler (KL) divergence, augmented by prediction-consistency trust scoring and latency-aware edge assignment to jointly mitigate data heterogeneity, device unreliability, and communication constraints. Second, it employs a resource-aware dynamic model splitting strategy to adaptively partition the LLM into three segments across clients and edge servers, with the cloud used only for adapter aggregation, enabling an effective balance between on-device computation cost and global convergence stability. Third, it incorporates a lightweight communication scheme based on computational sketches combined with semantic subspace orthogonal perturbation (SS-OP) to reduce communication overhead while mitigating privacy leakage during model exchanges across the network. Extensive experiments across diverse NLP tasks demonstrate that ELSA consistently outperforms state-of-the-art baselines in terms of adaptability, convergence behavior, and robustness, establishing a scalable and privacy-aware solution for edge-side LLM fine-tuning under resource constraints.
format Preprint
id arxiv_https___arxiv_org_abs_2601_13824
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ELSA: Efficient LLM-Centric Split Aggregation for Privacy-Aware Hierarchical Federated Learning over the Network Edge
Yang, Xiaohong
Xie, Tong
Liwang, Minghui
Shang, Chikai
Lu, Yang
Jiao, Zhenzhen
Fu, Liqun
Hosseinalipour, Seyyedali
Machine Learning
Training large language models (LLMs) at the network edge faces fundamental challenges arising from device resource constraints, severe data heterogeneity, and heightened privacy risks. To address these challenges, we propose ELSA (Efficient LLM-centric Split Aggregation), a novel framework that systematically integrates split learning (SL) and hierarchical federated learning (HFL) for distributed LLM fine-tuning over resource-constrained edge networks. ELSA introduces three key innovations. First, it employs a task-agnostic, behavior-aware client clustering mechanism that constructs semantic fingerprints using public probe inputs and symmetric Kullback-Leibler (KL) divergence, augmented by prediction-consistency trust scoring and latency-aware edge assignment to jointly mitigate data heterogeneity, device unreliability, and communication constraints. Second, it employs a resource-aware dynamic model splitting strategy to adaptively partition the LLM into three segments across clients and edge servers, with the cloud used only for adapter aggregation, enabling an effective balance between on-device computation cost and global convergence stability. Third, it incorporates a lightweight communication scheme based on computational sketches combined with semantic subspace orthogonal perturbation (SS-OP) to reduce communication overhead while mitigating privacy leakage during model exchanges across the network. Extensive experiments across diverse NLP tasks demonstrate that ELSA consistently outperforms state-of-the-art baselines in terms of adaptability, convergence behavior, and robustness, establishing a scalable and privacy-aware solution for edge-side LLM fine-tuning under resource constraints.
title ELSA: Efficient LLM-Centric Split Aggregation for Privacy-Aware Hierarchical Federated Learning over the Network Edge
topic Machine Learning
url https://arxiv.org/abs/2601.13824