Cited time in webofscience Cited time in scopus

Communication-Efficient and Drift-Robust Federated Learning via Elastic Net

Title
Communication-Efficient and Drift-Robust Federated Learning via Elastic Net
Author(s)
Seonhyeong Kim
DGIST Authors
Seonhyeong KimDaewon SeoYongjune Kim
Advisor
서대원
Co-Advisor(s)
Yongjune Kim
Issued Date
2023
Awarded Date
2023-02-01
Type
Thesis
Description
Federated learning, Machine learning, Optimization
Abstract
Federated learning (FL) is a distributed method to train a global model over a set of local clients
while keeping data localized. It reduces the risks of privacy and security but faces important challenges
including expensive communication costs and client drift issues. To address these issues, we propose
FedElasticNet, a communication-efficient and drift-robust FL framework leveraging the elastic net.
It repurposes two types of the elastic net regularizers (i.e., l1 and l2 penalties on the local model
updates): (1) the l1-norm regularizer sparsifies the local updates to reduce the communication costs
and (2) the l2-norm regularizer resolves the client drift problem by limiting the impact of drifting
local updates due to data heterogeneity. FedElasticNet is a general framework for FL; hence, without
additional costs, it can be integrated into prior FL techniques, e.g., FedAvg, FedProx, SCAFFOLD,
and FedDyn. We show that our framework effectively resolves the communication cost and client
drift problems simultaneously.; 연합 학습은 로컬 클라이언트가 가지고 있는 데이터를 공유하지 않으면서 로컬 클라이언트들의
데이터에 대한 글로벌 모델을 학습하는 분산 학습 방법이다. 연합 학습은 개인 정보 유출과 해
킹의 위험을 줄이지만 값비싼 통신 비용과 Client drift 문제를 포함한 중요한 해결 과제가 남아있
다. 이러한 문제를 해결하기 위해 Elastic Net 을 활용하는 통신 효율적이고 Client drift 에 강한 연
합학습 프레임워크인 FedElasticNet 을 제안한다. 이는 두 가지 유형의 Elastic Net Regularizer term
(즉, 로컬 모델 업데이트에 대한 ℓ1 및 ℓ2 페널티)의 목적을 변경한다. 본 논문에서 데이터 이질
성으로 인한 로컬 업데이트 드리프트의 영향을 제한하여 Client drift 문제를 해결한다. FedElasticNet 은 연합 학습에서 사용가능한 일반 프레임워크이다. 따라서 추가적인 수정없이 FedAvg,
FedProx, SCAFFOLD 및 FedDyn 과 같은 이전 연합 학습 기술에 통합할 수 있다. 본 논문에서 제
안하는 프레임워크가 통신 비용과 Client drift 문제를 동시에 효과적으로 해결한다는 것을 보여준
다.
Table Of Contents
Ⅰ. Introduction 1
Contributions 1
Ⅱ. Related Work 1
Related Work 2
Ⅲ. Proposed Method 2
FedElasticNet 3
3.1 3
3.2 3
3.3 4
Ⅳ. Experiments 5
Experimental Setup 6
Evaluation of Methods 6
Ⅴ. Conclusion 9
Ⅵ. Appendix 10
6.1 10
6.2 10
6.3 12
Ⅶ. Proof 14
URI
http://hdl.handle.net/20.500.11750/45748

http://dgist.dcollection.net/common/orgView/200000653511
DOI
10.22677/THESIS.200000653511
Degree
Master
Department
Department of Electrical Engineering and Computer Science
Publisher
DGIST
Related Researcher
  • 서대원 Seo, Daewon
  • Research Interests wireless communications; information theory; machine learning theorey
Files in This Item:

There are no files associated with this item.

Appears in Collections:
Department of Electrical Engineering and Computer Science Theses Master

qrcode

  • twitter
  • facebook
  • mendeley

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE