IEICE Technical Committee Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2022-01-21 11:20
A Study on Elucidating Hierarchical Neural Processing of Visual and Semantic Information using Deep Learning
Haruka Kawasaki (Ochadai), Satoshi Nishida (NICT), Ichiro Kobayashi (Ochadai) NC2021-32
Abstract (in Japanese) (See Japanese page) 
(in English) As an objective to explore the neural hierarchical processing underlying the transition from visual to semantic information, this study conducted encoding modeling using convolutional neural network (CNN) and language NN, which handle visual and verbal information, respectively. The encoding model predicted brain activity, evoked by movies, from the features of the movies extracted from each NN layer. By analyzing encoding models’ prediction in terms of the modality and hierarchy of the NNs, we investigated the cortical localization and representational content of visual and semantic neural information from the aspect of hierarchical processing. We found that the encoding models for the CNN higher layers and those for the language NN explained neural information in similar cortical regions but captured highly different representational content. Our finding suggests that although visual and semantic information shows more similar cortical localization as the visual information gets more complicated, the representational content of these information has a substantial gap, even in the same cortical region, which cannot be sufficiently bridged by the modeling with CNNs and language NNs alone.
Keyword (in Japanese) (See Japanese page) 
(in English) Brain activity / Hierachial processing / CNN / BERT / fMRI / Representational similarity analysis / /  
Reference Info. IEICE Tech. Rep., vol. 121, no. 338, NC2021-32, pp. 7-12, Jan. 2022.
Paper # NC2021-32 
Date of Issue 2022-01-14 (NC) 
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 NC2021-32

Conference Information
Committee NLP MICT MBE NC  
Conference Date 2022-01-21 - 2022-01-23 
Place (in Japanese) (See Japanese page) 
Place (in English) Online 
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To NC 
Conference Code 2022-01-NLP-MICT-MBE-NC 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) A Study on Elucidating Hierarchical Neural Processing of Visual and Semantic Information using Deep Learning 
Sub Title (in English)  
Keyword(1) Brain activity  
Keyword(2) Hierachial processing  
Keyword(3) CNN  
Keyword(4) BERT  
Keyword(5) fMRI  
Keyword(6) Representational similarity analysis  
Keyword(7)  
Keyword(8)  
1st Author's Name Haruka Kawasaki  
1st Author's Affiliation Ochanomizu University (Ochadai)
2nd Author's Name Satoshi Nishida  
2nd Author's Affiliation National Institute of Information and Communications Technology (NICT)
3rd Author's Name Ichiro Kobayashi  
3rd Author's Affiliation Ochanomizu University (Ochadai)
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 2022-01-21 11:20:00 
Presentation Time 25 minutes 
Registration for NC 
Paper # NC2021-32 
Volume (vol) vol.121 
Number (no) no.338 
Page pp.7-12 
#Pages
Date of Issue 2022-01-14 (NC) 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan