Saved in:
Bibliographic Details
Main Authors: Bartholet, Marc, Kim, Taehyeon, Beuret, Ami, Yun, Se-Young, Buhmann, Joachim M.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.06974
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909211552645120
author Bartholet, Marc
Kim, Taehyeon
Beuret, Ami
Yun, Se-Young
Buhmann, Joachim M.
author_facet Bartholet, Marc
Kim, Taehyeon
Beuret, Ami
Yun, Se-Young
Buhmann, Joachim M.
contents Federated Learning (FL) faces significant challenges with domain shifts in heterogeneous data, degrading performance. Traditional domain generalization aims to learn domain-invariant features, but the federated nature of model averaging often limits this due to its linear aggregation of local learning. To address this, we propose a robust framework, coined as hypernetwork-based Federated Fusion (hFedF), using hypernetworks for non-linear aggregation, facilitating generalization to unseen domains. Our method employs client-specific embeddings and gradient alignment techniques to manage domain generalization effectively. Evaluated in both zero-shot and few-shot settings, hFedF demonstrates superior performance in handling domain shifts. Comprehensive comparisons on PACS, Office-Home, and VLCS datasets show that hFedF consistently achieves the highest in-domain and out-of-domain accuracy with reliable predictions. Our study contributes significantly to the under-explored field of Federated Domain Generalization (FDG), setting a new benchmark for performance in this area.
format Preprint
id arxiv_https___arxiv_org_abs_2402_06974
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Hypernetwork-Driven Model Fusion for Federated Domain Generalization
Bartholet, Marc
Kim, Taehyeon
Beuret, Ami
Yun, Se-Young
Buhmann, Joachim M.
Machine Learning
Federated Learning (FL) faces significant challenges with domain shifts in heterogeneous data, degrading performance. Traditional domain generalization aims to learn domain-invariant features, but the federated nature of model averaging often limits this due to its linear aggregation of local learning. To address this, we propose a robust framework, coined as hypernetwork-based Federated Fusion (hFedF), using hypernetworks for non-linear aggregation, facilitating generalization to unseen domains. Our method employs client-specific embeddings and gradient alignment techniques to manage domain generalization effectively. Evaluated in both zero-shot and few-shot settings, hFedF demonstrates superior performance in handling domain shifts. Comprehensive comparisons on PACS, Office-Home, and VLCS datasets show that hFedF consistently achieves the highest in-domain and out-of-domain accuracy with reliable predictions. Our study contributes significantly to the under-explored field of Federated Domain Generalization (FDG), setting a new benchmark for performance in this area.
title Hypernetwork-Driven Model Fusion for Federated Domain Generalization
topic Machine Learning
url https://arxiv.org/abs/2402.06974