Inverse Distance Aggregation for Federated Learning with Non-IID Data

Yousef Yeganeh; Azade Farshad; Nassir Navab; Shadi Albarqouni

doi:10.1007/978-3-030-60548-3_15

Inverse Distance Aggregation for Federated Learning with Non-IID Data

Yousef Yeganeh^*
, Azade Farshad
, Nassir Navab
, Shadi Albarqouni

^*Corresponding author for this work

Not published at Aalto University

Technical University of Munich
Johns Hopkins University
Imperial College London

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

91 Citations (Scopus)

Abstract

Federated learning (FL) has been a promising approach in the field of medical imaging in recent years. A critical problem in FL, specifically in medical scenarios is to have a more accurate shared model which is robust to noisy and out-of distribution clients. In this work, we tackle the problem of statistical heterogeneity in data for FL which is highly plausible in medical data where for example the data comes from different sites with different scanner settings. We propose IDA (Inverse Distance Aggregation), a novel adaptive weighting approach for clients based on meta-information which handles unbalanced and non-iid data. We extensively analyze and evaluate our method against the well-known FL approach, Federated Averaging as a baseline.

Original language	English
Title of host publication	Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings
Editors	Shadi Albarqouni, Spyridon Bakas, Konstantinos Kamnitsas, M. Jorge Cardoso, Bennett Landman, Wenqi Li, Fausto Milletari, Nicola Rieke, Holger Roth, Daguang Xu, Ziyue Xu
Publisher	Springer
Pages	150-159
Number of pages	10
ISBN (Print)	9783030605476
DOIs	https://doi.org/10.1007/978-3-030-60548-3_15
Publication status	Published - 2020
MoE publication type	A4 Conference publication
Event	International Workshop on Domain Adaptation and Representation Transfer - Lima, Peru Duration: 4 Oct 2020 → 8 Oct 2020 Conference number: 2

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12444 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Workshop

Workshop	International Workshop on Domain Adaptation and Representation Transfer
Abbreviated title	DART
Country/Territory	Peru
City	Lima
Period	04/10/2020 → 08/10/2020

Keywords

Deep learning
Distributed learning
Federated learning
Heterogeneous data
Privacy-preserving
Robustness

Access to Document

10.1007/978-3-030-60548-3_15

Cite this

Yeganeh, Y., Farshad, A., Navab, N., & Albarqouni, S. (2020). Inverse Distance Aggregation for Federated Learning with Non-IID Data. In S. Albarqouni, S. Bakas, K. Kamnitsas, M. J. Cardoso, B. Landman, W. Li, F. Milletari, N. Rieke, H. Roth, D. Xu, & Z. Xu (Eds.), Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings (pp. 150-159). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12444 LNCS). Springer. https://doi.org/10.1007/978-3-030-60548-3_15

Yeganeh, Yousef ; Farshad, Azade ; Navab, Nassir et al. / Inverse Distance Aggregation for Federated Learning with Non-IID Data. Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings. editor / Shadi Albarqouni ; Spyridon Bakas ; Konstantinos Kamnitsas ; M. Jorge Cardoso ; Bennett Landman ; Wenqi Li ; Fausto Milletari ; Nicola Rieke ; Holger Roth ; Daguang Xu ; Ziyue Xu. Springer, 2020. pp. 150-159 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{89a4b9f2b7534020846f8db7d227f9e4,

title = "Inverse Distance Aggregation for Federated Learning with Non-IID Data",

abstract = "Federated learning (FL) has been a promising approach in the field of medical imaging in recent years. A critical problem in FL, specifically in medical scenarios is to have a more accurate shared model which is robust to noisy and out-of distribution clients. In this work, we tackle the problem of statistical heterogeneity in data for FL which is highly plausible in medical data where for example the data comes from different sites with different scanner settings. We propose IDA (Inverse Distance Aggregation), a novel adaptive weighting approach for clients based on meta-information which handles unbalanced and non-iid data. We extensively analyze and evaluate our method against the well-known FL approach, Federated Averaging as a baseline.",

keywords = "Deep learning, Distributed learning, Federated learning, Heterogeneous data, Privacy-preserving, Robustness",

author = "Yousef Yeganeh and Azade Farshad and Nassir Navab and Shadi Albarqouni",

note = "Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG.; International Workshop on Domain Adaptation and Representation Transfer, DART ; Conference date: 04-10-2020 Through 08-10-2020",

year = "2020",

doi = "10.1007/978-3-030-60548-3\_15",

language = "English",

isbn = "9783030605476",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "150--159",

editor = "Shadi Albarqouni and Spyridon Bakas and Konstantinos Kamnitsas and Cardoso, \{M. Jorge\} and Bennett Landman and Wenqi Li and Fausto Milletari and Nicola Rieke and Holger Roth and Daguang Xu and Ziyue Xu",

booktitle = "Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings",

address = "Germany",

}

Yeganeh, Y, Farshad, A, Navab, N & Albarqouni, S 2020, Inverse Distance Aggregation for Federated Learning with Non-IID Data. in S Albarqouni, S Bakas, K Kamnitsas, MJ Cardoso, B Landman, W Li, F Milletari, N Rieke, H Roth, D Xu & Z Xu (eds), Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12444 LNCS, Springer, pp. 150-159, International Workshop on Domain Adaptation and Representation Transfer, Lima, Peru, 04/10/2020. https://doi.org/10.1007/978-3-030-60548-3_15

Inverse Distance Aggregation for Federated Learning with Non-IID Data. / Yeganeh, Yousef; Farshad, Azade; Navab, Nassir et al.
Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings. ed. / Shadi Albarqouni; Spyridon Bakas; Konstantinos Kamnitsas; M. Jorge Cardoso; Bennett Landman; Wenqi Li; Fausto Milletari; Nicola Rieke; Holger Roth; Daguang Xu; Ziyue Xu. Springer, 2020. p. 150-159 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12444 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Inverse Distance Aggregation for Federated Learning with Non-IID Data

AU - Yeganeh, Yousef

AU - Farshad, Azade

AU - Navab, Nassir

AU - Albarqouni, Shadi

N1 - Conference code: 2

PY - 2020

Y1 - 2020

N2 - Federated learning (FL) has been a promising approach in the field of medical imaging in recent years. A critical problem in FL, specifically in medical scenarios is to have a more accurate shared model which is robust to noisy and out-of distribution clients. In this work, we tackle the problem of statistical heterogeneity in data for FL which is highly plausible in medical data where for example the data comes from different sites with different scanner settings. We propose IDA (Inverse Distance Aggregation), a novel adaptive weighting approach for clients based on meta-information which handles unbalanced and non-iid data. We extensively analyze and evaluate our method against the well-known FL approach, Federated Averaging as a baseline.

AB - Federated learning (FL) has been a promising approach in the field of medical imaging in recent years. A critical problem in FL, specifically in medical scenarios is to have a more accurate shared model which is robust to noisy and out-of distribution clients. In this work, we tackle the problem of statistical heterogeneity in data for FL which is highly plausible in medical data where for example the data comes from different sites with different scanner settings. We propose IDA (Inverse Distance Aggregation), a novel adaptive weighting approach for clients based on meta-information which handles unbalanced and non-iid data. We extensively analyze and evaluate our method against the well-known FL approach, Federated Averaging as a baseline.

KW - Deep learning

KW - Distributed learning

KW - Federated learning

KW - Heterogeneous data

KW - Privacy-preserving

KW - Robustness

U2 - 10.1007/978-3-030-60548-3_15

DO - 10.1007/978-3-030-60548-3_15

M3 - Conference article in proceedings

AN - SCOPUS:85092203932

SN - 9783030605476

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 150

EP - 159

BT - Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings

A2 - Albarqouni, Shadi

A2 - Bakas, Spyridon

A2 - Kamnitsas, Konstantinos

A2 - Cardoso, M. Jorge

A2 - Landman, Bennett

A2 - Li, Wenqi

A2 - Milletari, Fausto

A2 - Rieke, Nicola

A2 - Roth, Holger

A2 - Xu, Daguang

A2 - Xu, Ziyue

PB - Springer

T2 - International Workshop on Domain Adaptation and Representation Transfer

Y2 - 4 October 2020 through 8 October 2020

ER -

Yeganeh Y, Farshad A, Navab N, Albarqouni S. Inverse Distance Aggregation for Federated Learning with Non-IID Data. In Albarqouni S, Bakas S, Kamnitsas K, Cardoso MJ, Landman B, Li W, Milletari F, Rieke N, Roth H, Xu D, Xu Z, editors, Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings. Springer. 2020. p. 150-159. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-60548-3_15

Inverse Distance Aggregation for Federated Learning with Non-IID Data

Abstract

Publication series

Workshop

Keywords

Access to Document

Fingerprint

Cite this