When More Parameters Hurt: Foundation Model Priors Amplify Worst-Client Disparity Under Extreme Federated Heterogeneity

• URL: http://arxiv.org/abs/2605.08992v1
• Date: Sat, 09 May 2026 15:22:08 GMT
• Title: When More Parameters Hurt: Foundation Model Priors Amplify Worst-Client Disparity Under Extreme Federated Heterogeneity
• Abstract summary: Federated learning (FL) is increasingly used to fine-tune foundation models (FMs) on distributed private data.<n>We compare worst-client accuracy between TextCNN (2.7M parameters) and DistilBERT with Low-Rank Adaptation (LoRA)<n>Under extreme label skew (alpha = 0.1), DistilBERT+LoRA produces a worst-client accuracy gap of 50.1% -- 56% larger than TextCNN's 32.2% gap, despite having 25x more parameters and extensive pretraining.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Federated learning (FL) is increasingly used to fine-tune foundation models (FMs) on distributed private data. The community largely assumes that large-scale pretraining serves as a 'rising tide that lifts all boats' in federated settings. However, our experiments reveal that these powerful priors can hinder rather than help the most disadvantaged clients under extreme heterogeneity. Through controlled experiments on federated text classification, we compare worst-client accuracy between TextCNN (2.7M parameters) and DistilBERT with Low-Rank Adaptation (LoRA, 66M parameters) across four Non-IID heterogeneity levels. Under extreme label skew (alpha = 0.1), DistilBERT+LoRA produces a worst-client accuracy gap of 50.1% -- 56% larger than TextCNN's 32.2% gap, despite having 25x more parameters and extensive pretraining. Under moderate heterogeneity (alpha >= 0.5), the pattern reverses: the FM nearly eliminates the gap. We call this the FM Fairness Paradox. We further show that an inverse-weighted LoRA aggregation method (FedAvgW) does not resolve the disparity, suggesting aggregation reweighting alone may be insufficient. Our results highlight the need for mechanisms that explicitly protect minority clients before deploying foundation models in high-stakes federated contexts such as healthcare and education.

Related papers

• FedProIn: Mitigating Client Drift for Learnable Prototypes in Federated Medical Imaging [3.6868085124383625]
  We propose a prototype-based, influence-aware federated learning framework (FedProIn)<n>FedProIn uses multiple learnable class prototypes to capture shared semantic structures across heterogeneous clients.<n> Experimental results on two publicly available medical datasets, HAM10000 and Matek-19, demonstrate that FedProIn achieves accuracies of 83.5% IID, 81.1% non-IID, 96.2% IID, 95.8% non-IID, respectively.
  arXiv  Detail & Related papers  (2026-07-05T07:54:17Z)
• Subspace-Constrained Federated Learning with Low-Rank Adaptation [0.0]
  Federated low-rank adaptation methods are attractive for fine-tuning large models under communication and privacy constraints.<n>We study whether this subspace misalignment leads to destructive aggregation and slower convergence in LoRA-based federated learning.<n>We propose a subspace-regularized federated LoRA objective that encourages local client updates to remain close to a shared global reference subspace.
  arXiv  Detail & Related papers  (2026-06-21T23:49:22Z)
• Stabilized Fine-Tuning with LoRA in Federated Learning: Mitigating the Side Effect of Client Size and Rank via the Scaling Factor [8.51407618718798]
  In distributed scenarios where privacy constraints necessitate Federated Learning (FL), the integration of Low-Rank Adaptation (LoRA) is often unstable.<n>This paper introduces Stabilized Federated LoRA (SFed-LoRA), a framework that theoretically characterizes the interaction between adapter rank and federated aggregation.<n>We demonstrate that SFed-LoRA prevents high-rank collapse, and achieves significantly improved stability and faster convergence.
  arXiv  Detail & Related papers  (2026-03-09T07:49:56Z)
• FedHL: Federated Learning for Heterogeneous Low-Rank Adaptation via Unbiased Aggregation [6.5370850242187855]
  Federated Learning (FL) facilitates the fine-tuning of Foundation Models (FMs) using distributed data sources.<n>Low-Rank Adaptation (LoRA) gaining popularity due to its low communication costs and strong performance.<n>Existing methods lack formal convergence guarantees due to parameter truncation and biased gradient updates.
  arXiv  Detail & Related papers  (2025-05-24T04:12:12Z)
• Project-Probe-Aggregate: Efficient Fine-Tuning for Group Robustness [61.45587642780908]
  We propose a three-step approach for parameter-efficient fine-tuning of image-text foundation models.<n>Our method improves its two key components: minority samples identification and the robust training algorithm.<n>Our theoretical analysis shows that our PPA enhances minority group identification and is Bayes optimal for minimizing the balanced group error.
  arXiv  Detail & Related papers  (2025-03-12T15:46:12Z)
• Federated Learning under Partially Class-Disjoint Data via Manifold Reshaping [64.58402571292723]
  We propose a manifold reshaping approach called FedMR to calibrate the feature space of local training. We conduct extensive experiments on a range of datasets to demonstrate that our FedMR achieves much higher accuracy and better communication efficiency.
  arXiv  Detail & Related papers  (2024-05-29T10:56:13Z)
• Heterogeneous LoRA for Federated Fine-tuning of On-Device Foundation Models [20.707283766914017]
  HetLoRA allows heterogeneous ranks across client devices and efficiently aggregates and distributes these heterogeneous LoRA modules. HetLoRA achieves improved convergence speed and final performance compared to homogeneous LoRA.
  arXiv  Detail & Related papers  (2024-01-12T07:52:07Z)
• Marginal Debiased Network for Fair Visual Recognition [59.05212866862219]
  We propose a novel marginal debiased network (MDN) to learn debiased representations. Our MDN can achieve a remarkable performance on under-represented samples.
  arXiv  Detail & Related papers  (2024-01-04T08:57:09Z)
• Fairness-enhancing mixed effects deep learning improves fairness on in- and out-of-distribution clustered (non-iid) data [6.596656267996196]
  We propose the Fair Mixed Effects Deep Learning (Fair MEDL) framework.<n>This framework quantifies cluster-invariant fixed effects (FE) and cluster-specific random effects (RE) through: 1) a cluster adversary for learning invariant FE, 2) a Bayesian neural network for RE, and 3) a mixing function combining FE and RE for final predictions.<n>Fair MEDL framework improves fairness by 86.4% for Age, 64.9% for Race, 57.8% for Sex, and 36.2% for Marital status, while maintaining robust predictive performance.
  arXiv  Detail & Related papers  (2023-10-04T20:18:45Z)
• DFedADMM: Dual Constraints Controlled Model Inconsistency for Decentralized Federated Learning [52.83811558753284]
  Decentralized learning (DFL) discards the central server and establishes a decentralized communication network. Existing DFL methods still suffer from two major challenges: local inconsistency and local overfitting.
  arXiv  Detail & Related papers  (2023-08-16T11:22:36Z)
• FedSkip: Combatting Statistical Heterogeneity with Federated Skip Aggregation [95.85026305874824]
  We introduce a data-driven approach called FedSkip to improve the client optima by periodically skipping federated averaging and scattering local models to the cross devices. We conduct extensive experiments on a range of datasets to demonstrate that FedSkip achieves much higher accuracy, better aggregation efficiency and competing communication efficiency.
  arXiv  Detail & Related papers  (2022-12-14T13:57:01Z)
• Beyond ADMM: A Unified Client-variance-reduced Adaptive Federated Learning Framework [82.36466358313025]
  We propose a primal-dual FL algorithm, termed FedVRA, that allows one to adaptively control the variance-reduction level and biasness of the global model. Experiments based on (semi-supervised) image classification tasks demonstrate superiority of FedVRA over the existing schemes.
  arXiv  Detail & Related papers  (2022-12-03T03:27:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.