• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Study of dynamic attentional selection mechanisms in deep neural networks

Research Project

Project/Area Number 16K16090
Research Category

Grant-in-Aid for Young Scientists (B)

Allocation TypeMulti-year Fund
Research Field Perceptual information processing
Research InstitutionTokyo Denki University

Principal Investigator

HIDAKA Akinori  東京電機大学, 理工学部, 准教授 (70553519)

Research Collaborator KURITA Takio  
Project Period (FY) 2016-04-01 – 2019-03-31
Project Status Completed (Fiscal Year 2018)
Budget Amount *help
¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
Fiscal Year 2018: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2017: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2016: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Keywords深層学習 / 選択的注意 / 畳み込みニューラルネットワーク / 顕著性マップ / 注意・注視 / ニューラルネットワーク / 画像認識 / 機械学習 / 情報工学
Outline of Final Research Achievements

For analyzing the mechanism of consecutive visual information processing which is observed in both of visual cortex of humans and deep convolutional neural networks (CNN), we proposed interlayer correlations analysis based on canonical correlation analysis. It can realize quantitative evaluation and visualization of similarity or difference between an arbitrary pair of internal layers in CNN.
By using CNN trained for predicting the location of attentional selection and human gaze, we conducted the experiments for quantitative comparison between the neural system in visual cortex and CNN. We compared internal responses of CNN which is trained for saliency map prediction and the responses of V1, V4, IT area when they see (or process) several natural images. As a result, we obtained several novel findings about similarity and difference between CNN and the neural system in visual cortex.

Academic Significance and Societal Importance of the Research Achievements

畳み込みニューラルネットワーク(CNN)の極めて強力な画像認識性能は,現在の人工知能ブームを巻き起こす直接的なきっかけであり,今なお中心的な原動力であり続けている.本研究では,CNNの従来研究では長らく考慮されてこなかった大脳視覚野の認知処理過程における「注意(Attention)」の機構に焦点を当て,CNNの内部反応と大脳視覚皮質との類似性や相違性を定量的に分析し,新たな知見を得た.これらは,強力な認識性能を持つが静的な振る舞いしかできないCNNに対し,ヒトの視覚認知機構における「注意」に基づく動的なフィードバック機構等を組み込んで柔軟な認知能力を実現しようとする際に,重要な知見となる.

Report

(4 results)
  • 2018 Annual Research Report   Final Research Report ( PDF )
  • 2017 Research-status Report
  • 2016 Research-status Report
  • Research Products

    (5 results)

All 2019 2018 2017

All Journal Article (1 results) (of which Peer Reviewed: 1 results,  Open Access: 1 results) Presentation (4 results) (of which Int'l Joint Research: 1 results)

  • [Journal Article] Data Visualization for Deep Neural Networks Based on Interlayer Canonical Correlation Analysis2018

    • Author(s)
      Akinori Hidaka, Takio Kurita
    • Journal Title

      Transactions of the Institute of Systems, Control and Information Engineers

      Volume: 31 Issue: 1 Pages: 10-20

    • DOI

      10.5687/iscie.31.10

    • NAID

      130006707945

    • ISSN
      1342-5668, 2185-811X
    • Related Report
      2017 Research-status Report
    • Peer Reviewed / Open Access
  • [Presentation] サル視覚皮質と深層畳み込みニューラルネットワークから獲得する saliency map モデルの画像情報表現2019

    • Author(s)
      我妻伸彦、日高章理、田村弘
    • Organizer
      日本視覚学会2019年冬季大会
    • Related Report
      2018 Annual Research Report
  • [Presentation] The correspondence between monkey visual areas and DCNN saliency map model for representations of natural images2019

    • Author(s)
      Nobuhiko Wagatsuma, Akinori Hidaka, Hiroshi Tamura
    • Organizer
      The 15th Asia-Pacific Conference on Vision
    • Related Report
      2018 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 悪条件下におけるFaster R-CNN の物体検出性能の検証と改善2017

    • Author(s)
      長峯脩兵,湧井力,日高章理,狩野弘之
    • Organizer
      第49回計測自動制御学会北海道支部学術講演会
    • Place of Presentation
      北海道大学(北海道札幌市)
    • Year and Date
      2017-02-22
    • Related Report
      2016 Research-status Report
  • [Presentation] 深層畳み込みニューラルネットワークが獲得する注意選択モデルに対する心理物理学的タスクの影響2017

    • Author(s)
      我妻伸彦,日高章理
    • Organizer
      2017年度 人工知能学会全国大会(第31回)
    • Related Report
      2017 Research-status Report 2016 Research-status Report

URL: 

Published: 2016-04-21   Modified: 2020-03-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi