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

Multimodal Adaptive Structural Deep Learning using Knowledge Representation

Research Project

Project/Area Number 21K17809
Research Category

Grant-in-Aid for Early-Career Scientists

Allocation TypeMulti-year Fund
Review Section Basic Section 61030:Intelligent informatics-related
Research InstitutionHiroshima City University (2022-2024)
Prefectural University of Hiroshima (2021)

Principal Investigator

Shin Kamada  広島市立大学, 情報科学研究科, 准教授 (30845178)

Project Period (FY) 2021-04-01 – 2025-03-31
Project Status Completed (Fiscal Year 2024)
Budget Amount *help
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2023: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2022: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2021: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
Keywords深層学習 / マルチモーダル / 構造適応型学習 / ビッグデータ / 知識獲得
Outline of Research at the Start

本研究では,音声,自然言語等の複数のモダリティを並列的に処理し,モダリティ間の関係性を考慮した上で合成・融合し,最終的な出力判定を行うマルチモーダル構造適応型深層学習を探求する。複数のモデル間の確率分布の違いをKL情報量により測定し,類似性に基づいてモダリティ間の共通成分を知識獲得手法について研究する。入力に欠損値があった場合や一つのモダリティで曖昧な判定が行われた場合でも,別のモダリティがもつ知識に基づいて補間し,その他の情報と合成することで人間のような高次の情報処理を実現する。動画等のビッグデータに適用し評価する。

Outline of Final Research Achievements

We developed an adaptive structural deep learning method which automatically generates/deletes the suitable number of hidden neurons and layers for given input data. In this research, we developed a multimodal deep learning model using multiple input modalities such as video and audio to improve accuracy. For the emotion recognition dataset with video and audio, the proposed model showed higher classification accuracy than the unimodal models.

Academic Significance and Societal Importance of the Research Achievements

深層学習は,画像認識を中心として発展してきたが,近年では,テキストや音声認識等にみられるように,複数のモダリティを扱う深層学習法の開発が進んでいる。本研究のように,複数モダリティの互いの影響を質的・量的に観測した上で,適切な合成・融合・変換を行い,最終的な判定を行う仕組みを開発できれば,より性能が向上し,人工知能の研究がさらなる進化を遂げると考えられる。

Report

(5 results)
  • 2024 Annual Research Report   Final Research Report ( PDF )
  • 2023 Research-status Report
  • 2022 Research-status Report
  • 2021 Research-status Report
  • Research Products

    (16 results)

All 2025 2024 2023 2022 2021

All Journal Article (5 results) (of which Peer Reviewed: 5 results,  Open Access: 2 results) Presentation (11 results) (of which Int'l Joint Research: 3 results,  Invited: 1 results)

  • [Journal Article] Teacher-Student構造適応型深層学習によるセグメンテーションと送電鉄塔外観点検画像劣化診断への適用2025

    • Author(s)
      市村匠,鎌田真,山口亮,田中耕一
    • Journal Title

      計測自動制御学会論文集

      Volume: Vol.61, No.5

    • Related Report
      2024 Annual Research Report
    • Peer Reviewed
  • [Journal Article] 3D Lung Tumor Segmentation System using Adaptive Structural Deep Belief Network2024

    • Author(s)
      Shin Kamada, Takumi Ichimura
    • Journal Title

      Intelligent Systems Reference Library: Advances in Intelligent Disease Diagnosis and Treatment

      Volume: to appear in 2024

    • Related Report
      2023 Research-status Report
    • Peer Reviewed
  • [Journal Article] Automatic Extraction of Road Networks by Using Teacher-Student Adaptive Structural Deep Belief Network and Its Application to Landslide Disaster2023

    • Author(s)
      Shin Kamada, Takumi Ichimura
    • Journal Title

      IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

      Volume: 16 Pages: 6310-6324

    • DOI

      10.1109/jstars.2023.3293593

    • Related Report
      2023 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] A Teacher-Student based Adaptive Structural Deep learning Model and Its Estimating Uncertainty of Image Data2023

    • Author(s)
      Takumi Ichimura, Shin Kamada, Toshihide Harada and Ken Inoue
    • Journal Title

      Handbook of Statistics Volume 49: Artificial Intelligence

      Volume: 49

    • Related Report
      2022 Research-status Report
    • Peer Reviewed
  • [Journal Article] An Ensemble Learning Method of Adaptive Structural Deep Belief Network for AffectNet2022

    • Author(s)
      Takumi Ichimura, Shin Kamada
    • Journal Title

      International Journal of Smart Computing and Artificial Intelligence

      Volume: 6 Issue: 1 Pages: 1

    • DOI

      10.52731/ijscai.v6.i1.640

    • ISSN
      2185-9906, 2185-9914
    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Presentation] 顔表情画像とAction Unitを用いたVision TransformerによるRAVDESSデータセットの感情分類2025

    • Author(s)
      鎌田真,市村匠
    • Organizer
      計測自動制御学会第52回知能システムシンポジウム,pp.112-117
    • Related Report
      2024 Annual Research Report
  • [Presentation] Multimodal Adaptive Structural Deep Belief Network for Emotion Recognition on RAVDESS Dataset2024

    • Author(s)
      Shin Kamada, Takumi Ichimura
    • Organizer
      Proc. of 2024 16th International Congress on Advanced Applied Informatics (IIAI-AAI), pp.292-298
    • Related Report
      2024 Annual Research Report
    • Int'l Joint Research
  • [Presentation] マルチモーダル構造適応型深層学習によるRVDESSデータセットの感情分類2024

    • Author(s)
      鎌田真,市村匠
    • Organizer
      計測自動制御学会第51回知能システムシンポジウム,pp.152-157
    • Related Report
      2023 Research-status Report
  • [Presentation] A Segmentation Method of Lung Tumor by using Adaptive Structural Deep Belief Network2023

    • Author(s)
      Shin Kamada, Takumi Ichimura
    • Organizer
      Proc. of The SICE Annual Conference 2023 (SICE 2023), pp.1529-1534
    • Related Report
      2023 Research-status Report
    • Int'l Joint Research
  • [Presentation] 構造適応型深層学習を用いた肺腫瘍セグメンテーション手法の開発2022

    • Author(s)
      鎌田真,市村匠,河原大輔
    • Organizer
      計測自動制御学会第21回コンピューテーショナル・インテリジェンス研究会,pp.1-5
    • Related Report
      2022 Research-status Report
  • [Presentation] Teacher-Student型構造適応型深層学習モデルにおける複数GPU計算機の自動計算2022

    • Author(s)
      市村匠,鎌田真
    • Organizer
      2022 IEEE SMC Hiroshima Chapter Young Researchers WorkShop,pp.69-73
    • Related Report
      2022 Research-status Report
  • [Presentation] Automatic Extraction of Road Networks from Satellite Images by using Adaptive Structural Deep Belief Network2021

    • Author(s)
      Shin Kamada, Takumi Ichimura
    • Organizer
      Proc. of 10th International Conference on Informatics, Electronics & Vision (ICIEV 2021), paper 37
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] 構造適応型深層学習を用いた道路網認識手法RoadTracerによる土砂検出の試み2021

    • Author(s)
      鎌田真,市村匠
    • Organizer
      インテリジェント・システム・シンポジウム2021 (FAN2021),pp.181-186
    • Related Report
      2021 Research-status Report
  • [Presentation] Teacher-Student構造適応型深層学習によるアンサンブル学習と認知症MRI画像診断への応用2021

    • Author(s)
      鎌田真,市村匠,原田俊英
    • Organizer
      2021 IEEE SMC Hiroshima Chapter Young Researchers WorkShop,pp.30-35
    • Related Report
      2021 Research-status Report
  • [Presentation] 構造適応型深層学習法によるRoadTracerの道路検出に対する一考察2021

    • Author(s)
      市村匠,鎌田真
    • Organizer
      2021 IEEE SMC Hiroshima Chapter Young Researchers WorkShop,pp.16-23
    • Related Report
      2021 Research-status Report
  • [Presentation] Teacher-Student構造適応型深層学習によるアンサンブル学習と認知症MRI画像診断への応用2021

    • Author(s)
      市村匠,鎌田真,原田俊英,井上健
    • Organizer
      第33回日本老年医学会 中国地方会, No.23
    • Related Report
      2021 Research-status Report
    • Invited

URL: 

Published: 2021-04-28   Modified: 2026-01-16  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi