Paper Abstract and Keywords |
Presentation |
2021-03-03 14:40
Semantic and Quantitative Explanation for Networks using Feature Interaction Bohui Xia, Xueting Wang, Toshihiko Yamasaki (The Univ. of Tokyo) IMQ2020-35 IE2020-75 MVE2020-67 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Deep learning-based methods have shown remarkable performances in many tasks. However, it is hard for us to interpret the models, hence explainability of networks have been explored actively. In this work, we propose a method to generate semantic and quantitative explanations using attribute interactions based on an existing model. More specifically, we consider not only contributions from each attribute but also their interactions to prevent low explanatory performance. We test our model on multiple datasets and demonstrate how our method works in real world prediction problems. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Deep neural networks / Explainability / Semantic attributes / Feature interaction / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 120, no. 391, MVE2020-67, pp. 127-132, March 2021. |
Paper # |
MVE2020-67 |
Date of Issue |
2021-02-22 (IMQ, IE, MVE) |
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 |
IMQ2020-35 IE2020-75 MVE2020-67 |
Conference Information |
Committee |
MVE IMQ IE CQ |
Conference Date |
2021-03-01 - 2021-03-03 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
Online |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
|
Paper Information |
Registration To |
MVE |
Conference Code |
2021-03-MVE-IMQ-IE-CQ |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
Semantic and Quantitative Explanation for Networks using Feature Interaction |
Sub Title (in English) |
|
Keyword(1) |
Deep neural networks |
Keyword(2) |
Explainability |
Keyword(3) |
Semantic attributes |
Keyword(4) |
Feature interaction |
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Bohui Xia |
1st Author's Affiliation |
The University of Tokyo (The Univ. of Tokyo) |
2nd Author's Name |
Xueting Wang |
2nd Author's Affiliation |
The University of Tokyo (The Univ. of Tokyo) |
3rd Author's Name |
Toshihiko Yamasaki |
3rd Author's Affiliation |
The University of Tokyo (The Univ. of Tokyo) |
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 |
2021-03-03 14:40:00 |
Presentation Time |
25 minutes |
Registration for |
MVE |
Paper # |
IMQ2020-35, IE2020-75, MVE2020-67 |
Volume (vol) |
vol.120 |
Number (no) |
no.389(IMQ), no.390(IE), no.391(MVE) |
Page |
pp.127-132 |
#Pages |
6 |
Date of Issue |
2021-02-22 (IMQ, IE, MVE) |
|