Paper Abstract and Keywords |
Presentation |
2020-08-28 10:55
Theoretical Analysis on Convergence Acceleration of Deep-Unfolded Gradient Descent Satoshi Takabe, Tadashi Wadayama (NITech) SIP2020-35 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Deep unfolding is a promising deep learning technique whose network architecture is based on existing iterative algorithms. By unfolding the recursive structure of an iterative algorithm and embedding trainable parameters, deep unfolding can learn the parameters, which results in improving the convergence performance such as convergence speed. The goal of this paper is to give a plausible interpretation of convergence acceleration based on theoretical analyses of deep-unfolded gradient descent (DUGD). As a result, we will introduce Chebyshev steps based on Chebyshev polynomials, which well reproduce learned step sizes in DUGD and improve the convergence rate of GD. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
deep learning / deep unfolding / gradient descent / convergence rate / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 120, no. 142, SIP2020-35, pp. 25-30, Aug. 2020. |
Paper # |
SIP2020-35 |
Date of Issue |
2020-08-20 (SIP) |
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 |
SIP2020-35 |
Conference Information |
Committee |
SIP |
Conference Date |
2020-08-27 - 2020-08-28 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
Online |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
|
Paper Information |
Registration To |
SIP |
Conference Code |
2020-08-SIP |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Theoretical Analysis on Convergence Acceleration of Deep-Unfolded Gradient Descent |
Sub Title (in English) |
|
Keyword(1) |
deep learning |
Keyword(2) |
deep unfolding |
Keyword(3) |
gradient descent |
Keyword(4) |
convergence rate |
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Satoshi Takabe |
1st Author's Affiliation |
Nagoya Institute of Technology (NITech) |
2nd Author's Name |
Tadashi Wadayama |
2nd Author's Affiliation |
Nagoya Institute of Technology (NITech) |
3rd Author's Name |
|
3rd Author's Affiliation |
() |
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 |
() |
21st Author's Name |
|
21st Author's Affiliation |
() |
22nd Author's Name |
|
22nd Author's Affiliation |
() |
23rd Author's Name |
|
23rd Author's Affiliation |
() |
24th Author's Name |
|
24th Author's Affiliation |
() |
25th Author's Name |
|
25th Author's Affiliation |
() |
26th Author's Name |
/ / |
26th Author's Affiliation |
()
() |
27th Author's Name |
/ / |
27th Author's Affiliation |
()
() |
28th Author's Name |
/ / |
28th Author's Affiliation |
()
() |
29th Author's Name |
/ / |
29th Author's Affiliation |
()
() |
30th Author's Name |
/ / |
30th Author's Affiliation |
()
() |
31st Author's Name |
/ / |
31st Author's Affiliation |
()
() |
32nd Author's Name |
/ / |
32nd Author's Affiliation |
()
() |
33rd Author's Name |
/ / |
33rd Author's Affiliation |
()
() |
34th Author's Name |
/ / |
34th Author's Affiliation |
()
() |
35th Author's Name |
/ / |
35th Author's Affiliation |
()
() |
36th Author's Name |
/ / |
36th Author's Affiliation |
()
() |
Speaker |
Author-1 |
Date Time |
2020-08-28 10:55:00 |
Presentation Time |
25 minutes |
Registration for |
SIP |
Paper # |
SIP2020-35 |
Volume (vol) |
vol.120 |
Number (no) |
no.142 |
Page |
pp.25-30 |
#Pages |
6 |
Date of Issue |
2020-08-20 (SIP) |