Paper Abstract and Keywords |
Presentation |
2023-03-02 14:30
[Poster Presentation]
A Study on Eliminating Malicious Node in Federated Learning Reon Akai, Minoru Kuribayashi, Nobuo Funabiki (Okayama Univ) EMM2022-84 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Federated learning (FL) has been proposed to aggregate deep learning models trained at each node in order to utilize privacy-sensitive data stored separately at each.
However, the presence of malicious nodes among the nodes can degrade the performance of the entire system.
In this study, we investigate a method to prevent performance degradation of the entire system by excluding malicious nodes in federated averaging (FedAvg).
The weight parameters trained at malicious nodes must differ significantly in their statistical characteristics compared to the weight parameters of the normal nodes.
Therefore, the proposed method excludes outliers from the weight parameters received from multiple nodes in the deep neural network (DNN) model.
Our simulations show that this process does not cause much degradation in learning efficiency.
Furthermore, we proposed a method for FedAvg to save the weight parameters each round of training, and to load the file and resume learning, so that the entire system can return to the state before the performance degradation. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Federated Learning / Deep Learning / Machine Learning / Federated Averaging / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 122, no. 412, EMM2022-84, pp. 89-94, March 2023. |
Paper # |
EMM2022-84 |
Date of Issue |
2023-02-23 (EMM) |
ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
EMM2022-84 |
Conference Information |
Committee |
EMM |
Conference Date |
2023-03-02 - 2023-03-03 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
Fukue culture hall |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
|
Paper Information |
Registration To |
EMM |
Conference Code |
2023-03-EMM |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
A Study on Eliminating Malicious Node in Federated Learning |
Sub Title (in English) |
|
Keyword(1) |
Federated Learning |
Keyword(2) |
Deep Learning |
Keyword(3) |
Machine Learning |
Keyword(4) |
Federated Averaging |
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Reon Akai |
1st Author's Affiliation |
Okayama University (Okayama Univ) |
2nd Author's Name |
Minoru Kuribayashi |
2nd Author's Affiliation |
Okayama University (Okayama Univ) |
3rd Author's Name |
Nobuo Funabiki |
3rd Author's Affiliation |
Okayama University (Okayama Univ) |
4th Author's Name |
|
4th Author's Affiliation |
() |
5th Author's Name |
|
5th Author's Affiliation |
() |
6th Author's Name |
|
6th Author's Affiliation |
() |
7th Author's Name |
|
7th Author's Affiliation |
() |
8th Author's Name |
|
8th Author's Affiliation |
() |
9th Author's Name |
|
9th Author's Affiliation |
() |
10th Author's Name |
|
10th Author's Affiliation |
() |
11th Author's Name |
|
11th Author's Affiliation |
() |
12th Author's Name |
|
12th Author's Affiliation |
() |
13th Author's Name |
|
13th Author's Affiliation |
() |
14th Author's Name |
|
14th Author's Affiliation |
() |
15th Author's Name |
|
15th Author's Affiliation |
() |
16th Author's Name |
|
16th Author's Affiliation |
() |
17th Author's Name |
|
17th Author's Affiliation |
() |
18th Author's Name |
|
18th Author's Affiliation |
() |
19th Author's Name |
|
19th Author's Affiliation |
() |
20th Author's Name |
|
20th Author's Affiliation |
() |
Speaker |
Author-1 |
Date Time |
2023-03-02 14:30:00 |
Presentation Time |
75 minutes |
Registration for |
EMM |
Paper # |
EMM2022-84 |
Volume (vol) |
vol.122 |
Number (no) |
no.412 |
Page |
pp.89-94 |
#Pages |
6 |
Date of Issue |
2023-02-23 (EMM) |
|