Parameterizing Federated Continual Learning for Reproducible Research

Bart Cox; Jeroen Galjaard; Aditya Shankar; Jérémie Decouchant; Lydia Y. Chen

doi:10.1007/978-3-031-74643-7_35

Parameterizing Federated Continual Learning for Reproducible Research

Bart Cox^*, Jeroen Galjaard, Aditya Shankar, Jérémie Decouchant, Lydia Y. Chen

^*Corresponding author for this work

Data-Intensive Systems

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Abstract

Federated Learning (FL) systems evolve in heterogeneous and ever-evolving environments that challenge their performance. Under real deployments, the learning tasks of clients can also evolve with time, which calls for the integration of methodologies such as Continual Learning (CL). To enable research reproducibility, we propose a set of experimental best practices that precisely capture and emulate complex learning scenarios. To the best of our knowledge, our framework, Freddie, is the first entirely configurable framework for Federated Continual Learning (FCL), and it can be seamlessly deployed on a large number of machines leveraging containerization and Kubernetes. We demonstrate the effectiveness of Freddie on two use cases, (i) large-scale concurrent FL on CIFAR100 and (ii) heterogeneous task sequence on FCL, which highlight unaddressed performance challenges in FCL scenarios.

Original language	English
Title of host publication	Machine Learning and Principles and Practice of Knowledge Discovery in Databases
Subtitle of host publication	International Workshops of ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Revised Selected Papers, Part V
Editors	Rosa Meo, Fabrizio Silvestri
Place of Publication	Cham
Publisher	Springer
Pages	478-486
Number of pages	9
ISBN (Electronic)	978-3-031-74643-7
ISBN (Print)	978-3-031-74642-0
DOIs	https://doi.org/10.1007/978-3-031-74643-7_35
Publication status	Published - 2025
Event	Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2023 - Turin, Italy Duration: 18 Sept 2023 → 22 Sept 2023 https://2023.ecmlpkdd.org/

Publication series

Name	Communications in Computer and Information Science
Publisher	Springer
Volume	2137 CCIS
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2023
Country/Territory	Italy
City	Turin
Period	18/09/23 → 22/09/23
Internet address	https://2023.ecmlpkdd.org/

Bibliographical note

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

federated continual learning
resource and data heterogeneity
reproducible research

Access to Document

10.1007/978-3-031-74643-7_35

Cite this

Cox, B., Galjaard, J., Shankar, A., Decouchant, J., & Chen, L. Y. (2025). Parameterizing Federated Continual Learning for Reproducible Research. In R. Meo, & F. Silvestri (Eds.), Machine Learning and Principles and Practice of Knowledge Discovery in Databases: International Workshops of ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Revised Selected Papers, Part V (pp. 478-486). (Communications in Computer and Information Science; Vol. 2137 CCIS). Springer. https://doi.org/10.1007/978-3-031-74643-7_35

Cox, Bart ; Galjaard, Jeroen ; Shankar, Aditya et al. / Parameterizing Federated Continual Learning for Reproducible Research. Machine Learning and Principles and Practice of Knowledge Discovery in Databases: International Workshops of ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Revised Selected Papers, Part V. editor / Rosa Meo ; Fabrizio Silvestri. Cham : Springer, 2025. pp. 478-486 (Communications in Computer and Information Science).

@inproceedings{5ea3b2e892cf4093a98c4b096f231bad,

title = "Parameterizing Federated Continual Learning for Reproducible Research",

abstract = "Federated Learning (FL) systems evolve in heterogeneous and ever-evolving environments that challenge their performance. Under real deployments, the learning tasks of clients can also evolve with time, which calls for the integration of methodologies such as Continual Learning (CL). To enable research reproducibility, we propose a set of experimental best practices that precisely capture and emulate complex learning scenarios. To the best of our knowledge, our framework, Freddie, is the first entirely configurable framework for Federated Continual Learning (FCL), and it can be seamlessly deployed on a large number of machines leveraging containerization and Kubernetes. We demonstrate the effectiveness of Freddie on two use cases, (i) large-scale concurrent FL on CIFAR100 and (ii) heterogeneous task sequence on FCL, which highlight unaddressed performance challenges in FCL scenarios.",

keywords = "federated continual learning, resource and data heterogeneity, reproducible research",

author = "Bart Cox and Jeroen Galjaard and Aditya Shankar and J{\'e}r{\'e}mie Decouchant and Chen, {Lydia Y.}",

note = "Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.; Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2023 ; Conference date: 18-09-2023 Through 22-09-2023",

year = "2025",

doi = "10.1007/978-3-031-74643-7_35",

language = "English",

isbn = "978-3-031-74642-0",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "478--486",

editor = "Rosa Meo and Fabrizio Silvestri",

booktitle = "Machine Learning and Principles and Practice of Knowledge Discovery in Databases",

url = "https://2023.ecmlpkdd.org/",

}

Cox, B , Galjaard, J , Shankar, A , Decouchant, J & Chen, LY 2025, Parameterizing Federated Continual Learning for Reproducible Research. in R Meo & F Silvestri (eds), Machine Learning and Principles and Practice of Knowledge Discovery in Databases: International Workshops of ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Revised Selected Papers, Part V. Communications in Computer and Information Science, vol. 2137 CCIS, Springer, Cham, pp. 478-486, Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2023, Turin, Italy, 18/09/23. https://doi.org/10.1007/978-3-031-74643-7_35

Parameterizing Federated Continual Learning for Reproducible Research. / Cox, Bart ; Galjaard, Jeroen ; Shankar, Aditya et al.
Machine Learning and Principles and Practice of Knowledge Discovery in Databases: International Workshops of ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Revised Selected Papers, Part V. ed. / Rosa Meo; Fabrizio Silvestri. Cham: Springer, 2025. p. 478-486 (Communications in Computer and Information Science; Vol. 2137 CCIS).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Parameterizing Federated Continual Learning for Reproducible Research

AU - Cox, Bart

AU - Galjaard, Jeroen

AU - Shankar, Aditya

AU - Decouchant, Jérémie

AU - Chen, Lydia Y.

N1 - Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2025

Y1 - 2025

N2 - Federated Learning (FL) systems evolve in heterogeneous and ever-evolving environments that challenge their performance. Under real deployments, the learning tasks of clients can also evolve with time, which calls for the integration of methodologies such as Continual Learning (CL). To enable research reproducibility, we propose a set of experimental best practices that precisely capture and emulate complex learning scenarios. To the best of our knowledge, our framework, Freddie, is the first entirely configurable framework for Federated Continual Learning (FCL), and it can be seamlessly deployed on a large number of machines leveraging containerization and Kubernetes. We demonstrate the effectiveness of Freddie on two use cases, (i) large-scale concurrent FL on CIFAR100 and (ii) heterogeneous task sequence on FCL, which highlight unaddressed performance challenges in FCL scenarios.

AB - Federated Learning (FL) systems evolve in heterogeneous and ever-evolving environments that challenge their performance. Under real deployments, the learning tasks of clients can also evolve with time, which calls for the integration of methodologies such as Continual Learning (CL). To enable research reproducibility, we propose a set of experimental best practices that precisely capture and emulate complex learning scenarios. To the best of our knowledge, our framework, Freddie, is the first entirely configurable framework for Federated Continual Learning (FCL), and it can be seamlessly deployed on a large number of machines leveraging containerization and Kubernetes. We demonstrate the effectiveness of Freddie on two use cases, (i) large-scale concurrent FL on CIFAR100 and (ii) heterogeneous task sequence on FCL, which highlight unaddressed performance challenges in FCL scenarios.

KW - federated continual learning

KW - resource and data heterogeneity

KW - reproducible research

UR - http://www.scopus.com/inward/record.url?scp=85215999139&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-74643-7_35

DO - 10.1007/978-3-031-74643-7_35

M3 - Conference contribution

AN - SCOPUS:85215999139

SN - 978-3-031-74642-0

T3 - Communications in Computer and Information Science

SP - 478

EP - 486

BT - Machine Learning and Principles and Practice of Knowledge Discovery in Databases

A2 - Meo, Rosa

A2 - Silvestri, Fabrizio

PB - Springer

CY - Cham

T2 - Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2023

Y2 - 18 September 2023 through 22 September 2023

ER -

Cox B , Galjaard J , Shankar A , Decouchant J , Chen LY. Parameterizing Federated Continual Learning for Reproducible Research. In Meo R, Silvestri F, editors, Machine Learning and Principles and Practice of Knowledge Discovery in Databases: International Workshops of ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Revised Selected Papers, Part V. Cham: Springer. 2025. p. 478-486. (Communications in Computer and Information Science). doi: 10.1007/978-3-031-74643-7_35