paper-with-me

홈 › Papers

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

2026-05-09 · Kiran Naseer, Umar Shoaib arxiv

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.

📄 PDF Abstract BibTeX arXiv:2605.08992

Code (0)

등록된 구현이 없습니다.

Tasks

Text ClassificationFederated Learning

Similar Papers 제목 키워드 기반

Geometry without Position? When Positional Embeddings Help and Hurt Spatial Reasoning

2026-01-29 · Jian Shi, Michael Birsak, Wenqing Cui, Zhenyu Li 외 arxiv

This paper revisits the role of positional embeddings (PEs) within vision transformers (ViTs) from a geometric perspective. We show that PEs are not mere token indices but effectively function as geometric priors that sh…

Spatial Reasoning

Beyond One-Size-Fits-All: Diagnosis-Driven Online Reinforcement Learning with Offline Priors

2026-06-24 · Guozheng Ma, Lu Li, Zilin Wang, Pierre-Luc Bacon 외 arxiv

Online reinforcement learning (RL) agents increasingly depend on knowledge acquired offline to achieve practical efficiency. Originally studied in offline-to-online RL, this paradigm now spans foundation model post-train…

Reinforcement Learning

Adapting Topic Models using Lexical Associations with Tree Priors

2017-09-01 · EMNLP 2017 9 · Weiwei Yang, Jordan Boyd-Graber, Philip Resnik

Models work best when they are optimized taking into account the evaluation criteria that people care about. For topic models, people often care about interpretability, which can be approximated using measures of lexical…

Machine TranslationTopic ModelsWord Embeddings

Controllable-LPMoE: Adapting to Challenging Object Segmentation via Dynamic Local Priors from Mixture-of-Experts

2025-10-24 · Yanguang Sun, Jiawei Lian, Jian Yang, Lei Luo arxiv

Large-scale foundation models provide powerful feature representations for downstream object segmentation tasks. However, when adapted to specific tasks through the full-parameter fine-tuning, the enormous parameters bei…

Object Segmentation

When Covariate-shifted Data Augmentation Increases Test Error And How to Fix It

2019-09-25 · Sang Michael Xie*, Aditi Raghunathan*, Fanny Yang, John C. Duchi 외

Empirically, data augmentation sometimes improves and sometimes hurts test error, even when only adding points with labels from the true conditional distribution that the hypothesis class is expressive enough to fit. In…

Data Augmentationregression