OTSUKA Kazuhiro

Affiliation

Faculty of Engineering, Division of Intelligent Systems Engineering

Job Title

Associate Professor

Research Fields, Keywords

Communication Science, Multimodal Interaction, Social Signal Processing



Education 【 display / non-display

  • 2005.4
    -
    2007.3

    Nagoya University   Doctor Course   Completed

  • 1993.4
    -
    1995.3

    Yokohama National University   Electrical and Computer Engineering   Master Course   Completed

  • 1989.4
    -
    1993.3

    Yokohama National University   Division of Electrical and Computer Enginnering   Graduated

Degree 【 display / non-display

  • Doctor of Information Science - Nagoya University

  • Master of Engineering - Yokohama National University

Campus Career 【 display / non-display

  • 2020.4
     
     

    Duty   Yokohama National UniversityFaculty of Engineering   Division of Intelligent Systems Engineering   Associate Professor  

External Career 【 display / non-display

  • 2012.10
    -
    2020.3

     

  • 2007.4
    -
    2012.9

     

  • 2003.4
    -
    2007.3

     

  • 2001.4
    -
    2003.3

     

  • 1999.12
    -
    2001.3

     

display all >>

Academic Society Affiliations 【 display / non-display

  • 2022.4
     
     
     

    The Japanese Society for Artificial Intelligence

  • 2005
     
     
     

    IEEE(Institute of Electrical and Electronics Engineering)

  • 1996
     
     
     

    電子情報通信学会

  • 1996
     
     
     

    情報処理学会

Research Areas 【 display / non-display

  • Informatics / Intelligent informatics

  • Informatics / Human interface and interaction

 

Thesis for a degree 【 display / non-display

  • 時系列画像に基づく動的事象の推測に関する研究

    大塚和弘

    2007.3

    Doctoral Thesis   Single Work  

Papers 【 display / non-display

  • Prediction of Participant's Subjective Impressions based on Head-Movement Functions in Multiparty Conversations

    OTSUCHI Shumpei, ISHII Yoko, NAKATANI Momoko, OTSUKA Kazuhiro

    J105-D ( 9 )   504 - 517   2022.9

    DOI CiNii Research

     More details

    Language:Japanese   Publishing type:Research paper (scientific journal)   Joint Work  

    This paper proposes novel prediction models for the subjective impressions of meeting participants, based on their head-movement functions. The goal is to estimate the self-rated impression scores on the impression items consisting of `atmosphere,' `enjoyment,' `willingness,' and `concentration,' which were rated every two-minute intervals in the meetings. The prediction model employs Random-Forest regressors that input head-movement function features based on 10 head-movement functions detected with convolutional neural networks using head pose sequences measured in the meeting videos. The experiment results show the models' predictability and reveal the relationship between the impressions and the participants' behaviors.

  • Recognizing Functional Interactions via Head Movements by Deep Transfer Learning in Multiparty Conversations

    MORI Takashi, OTSUKA Kazuhiro

    J105-A ( 10 )   111 - 124   2022.6  [Reviewed]

    DOI CiNii Research

     More details

    Language:Japanese   Publishing type:Research paper (scientific journal)   Publisher:The Institute of Electronics, Information and Communication Engineers   Single Work  

    In group meetings, the participant's head movements play various functions. This study extends the concept of individual head-movement function (HMF) to interactional functions between the speaker and listener in meetings. We propose strategies to transfer the existing CNN-based HMF recognition model into the interactional models that can recognize the interactional functions better than the simple logical-product decision of two individual CNN models. The transfer strategies include a distributed representation of multiple HMFs, called a functional spectrum vector extracted from the CNNs' middle layers. The experiments confirmed that the transferred models outperformed the logical-product-based model up to 13.9 points in F-score.

  • Explainable Models for Predicting Interlocuters' Subjective Impressions based on Nonverbal Functional Features

    OTSUCHI Shumpei, MIYOSHI Kazuki, ISHII Yoko, ISHII Ryo, EITOKU Shin-ichiro, OTSUKA Kazuhiro

    Proceedings of the Annual Conference of JSAI   JSAI2022 ( 0 )   3H3OS12a05 - 3H3OS12a05   2022

    DOI CiNii Research

     More details

    Language:Japanese   Publishing type:Research paper (scientific journal)   Publisher:The Japanese Society for Artificial Intelligence   Joint Work  

    <p>An explainable framework is proposed to predict the interlocutors' subjective impressions in group meetings. The goal is to explain when and what kind of nonverbal behaviors affected interlocutors' impressions during meetings. To that end, we formulate a two-fold framework consisting of the regression models of interlocutors' impression scores based on functional head-movement features, followed by the estimation of the temporal distribution of SHAP-based feature contribution, which is obtained with the kernel density estimation of the temporal occurrence probabilities of head-movement functions. The former stage identifies the behaviors related to the impressions, and the later stage suggests the timing of the behaviors, by locating the maximum point of the temporal feature contribution curve, based on an assumption that temporary intensive behaviors lead to form a strong impression. This report shows preliminary results and analyses applied to 4-party 17-group discussions.</p>

  • Recognizing Head-Movement Functions in Conversations by Using Feature Inflation-Deflation Networks

    TAKEDA Kazuki, OTSUKA Kazuhiro

    J105-A ( 2 )   26 - 37   2021.9  [Reviewed]

    DOI CiNii Research

     More details

    Language:Japanese   Publishing type:Research paper (scientific journal)   Publisher:The Institute of Electronics, Information and Communication Engineers   Joint Work  

    A new deep neural network module called an Inflation-Deflation (I/DeF) module is proposed to improve the recognition performance of the head-movement functions in group meetings, using Convolutional Neural Networks (CNNs). To facilitate the learning of the temporal features in the short windowed input head-movement sequence, the I/DeF module repeatedly cascades an inflation process and a deflation process. The former extends the input sequence in terms of its scale and temporal dimensions using a transposed convolution. The deflation process reduces and restores the data dimensions by using convolution. The experiments confirmed that I/DeF-integrated CNNs and other DNNs outperformed the existing DNNs.

  • Estimation of Personal Empathy Skill Level Using Dialogue Act and Eye-gaze during Turn-keeping/changing

    Ryo Ishii, Kazuhiro Otsuka, Shiro Kumano, Ryuichiro Higashinaka, Yushi Aono

    62 ( 1 )   100 - 114   2021.1  [Reviewed]

     More details

    Language:Japanese   Publishing type:Research paper (scientific journal)   Joint Work  

display all >>

Review Papers 【 display / non-display

  • 非言語情報が生み出すコミュニケーションのリアリティ

    大塚和弘

    NTT研究所発 触感コンテンツ専門誌ふるえ   8 ( 5 )   2016.12

     More details

    Language:Japanese   Publishing type:Article, review, commentary, editorial, etc. (other)   Single Work  

  • Prediction of "Who Will Be Next Speaker and When" in Multi-Party Meetings

    Ryo Ishii, Kazuhiro Otsuka, Shiro Kumano, Junji Yamato

    NTT Technical Review   13 ( 7 )   2015

     More details

    Language:English   Publishing type:Internal/External technical report, pre-print, etc.   Joint Work  

  • The Future of Idobata Kaigi (“Congregation at the Well”): Realizing Natural Conversations with Distance Access

    Kazuhiro Otsuka, Nobuhiro Furuyama, Mayumi Bono

    NII Today   48   6 - 7   2014.1

     More details

    Language:English   Publishing type:Internal/External technical report, pre-print, etc.   Joint Work  

  • 遠隔地間でも自然な会話を実現する、未来の井戸端会議とは

    大塚和弘,古山宣洋,坊農真弓

    国立情報学研究所 NII Today   62   6 - 7   2013.12

     More details

    Language:Japanese   Publishing type:Internal/External technical report, pre-print, etc.   Joint Work  

  • MM-Space: Recreating Multiparty Conversation Space by Using Dynamic Displays

    Kazuhiro Otsuka

    NTT Technical Review   10 ( 11 )   2012.11

     More details

    Language:English   Publishing type:Internal/External technical report, pre-print, etc.   Single Work  

display all >>

Awards 【 display / non-display

  • HCS研究会賞

    2022.1   電子情報通信学会 ヒューマンコミュニケーション基礎研究専門委員会   「対話中の非言語行動の機能スペクトラム解析に向けて〜頭部運動を対象とした機能コーパスと深層学習モデルの構築事例〜」

    Individual or group name of awards:大塚和弘

  • 特集テーマセッション賞

    2021.12   電子情報通信学会 HCGシンポジウム2021   複数人対話における深層転移学習による頭部運動相互作用の機能認識

    Individual or group name of awards:森 駿志, 大塚和弘

  • 情報処理学会論文誌ジャーナル特選論文賞

    2021.1   情報処理学会   話者継続・交替時における対話行為と視線行動に基づく共感スキルの推定

    Individual or group name of awards:石井 亮, 大塚和弘, 熊野史朗, 東中竜一郎, 青野裕司

  • 電子情報通信学会 HCGシンポジウム2018 特集テーマセッション賞

    2018.12   電子情報通信学会  

    Individual or group name of awards:石井 亮,大塚和弘,熊野史朗,東中竜一郎,富田準二

  • 電子情報通信学会 HCGシンポジウム2015 優秀インタラクティブ発表賞

    2015.12   電子情報通信学会  

    Individual or group name of awards:熊野史朗,大塚和弘,石井亮,大和淳司

display all >>

Presentations 【 display / non-display

  • 対話中の頭部運動機能の解釈の曖昧性を推定する深層回帰モデルの検討

    武田 一輝, 大塚 和弘

    第21回情報科学技術フォーラム(FIT2022)  

     More details

    Event date: 2022.9

    Language:Japanese   Presentation type:Oral presentation (general)  

  • 複数人対話における深層転移学習による頭部運動相互作用の機能認識

    森 駿志, 大塚和弘

    電子情報通信学会HCGシンポジウム2021   2021.12  電子情報通信学会

     More details

    Event date: 2021.12

    Language:Japanese   Presentation type:Oral presentation (general)  

  • Prediction of Interlocutors’ Subjective Impressions Based on Functional Head-Movement Features in Group Meetings

    Shumepi Otsuchi, Yoko Ishii, Momoko Nakatani, Kazuhiro Otsuka

    23rd ACM International Conference on Multimodal Interaction (ICMI2021)  2021.10  ACM

    DOI

     More details

    Event date: 2021.10

    Language:English   Presentation type:Poster presentation  

    Venue:Montreal, Canada   Country:Canada  

  • Deep Transfer Learning for Recognizing Functional Interactions via Head Movements in Multiparty Conversations

    Takashi Mori, Kazuhiro Otsuka

    23rd ACM International Conference on Multimodal Interaction (ICMI2021)  2021.10  ACM

    DOI

     More details

    Event date: 2021.10

    Language:English   Presentation type:Poster presentation  

    Venue:Montreal, Canada   Country:Canada  

  • Inflation-Deflation Networks for Recognizing Head-Movement Functions in Face-to-Face Conversations

    Kazuki Takeda, Kazuhiro Otsuka

    23rd ACM International Conference on Multimodal Interaction (ICMI2021)  2021.10  ACM

    DOI

     More details

    Event date: 2021.10

    Language:English   Presentation type:Poster presentation  

    Venue:Montreal, Canada   Country:Canada  

display all >>

 

Charge of on-campus class subject 【 display / non-display

  • 2022   Colloquium in Electrical and Computer Engineering II

    Graduate school of Engineering Science

  • 2022   Colloquium in Electrical and Computer Engineering I

    Graduate school of Engineering Science

  • 2022   Advanced information system I-S

    Graduate school of Engineering Science

  • 2022   Advanced information system I-F

    Graduate school of Engineering Science

  • 2022   Advanced information system Ⅱ-S

    Graduate school of Engineering Science

display all >>

Charge of off-campus class subject 【 display / non-display

  • 画像ハンドリング工学

 

Committee Memberships 【 display / non-display

  • 情報処理学会 情報規格調査会

    2018.7 - 2020.3  専門委員

     More details

    Committee type:Academic society 

  • 電子情報通信学会 ヴァーバル・ノンヴァーバル・コミュニケーション研究会

    2016.4 - 2020.3  委員長

     More details

    Committee type:Academic society 

  • 電子情報通信学会 和文論文誌D編集委員会

    2010.5 - 2014.6  編集委員

     More details

    Committee type:Academic society 

  • 電子情報通信学会 ヴァーバル・ノンヴァーバル・コミュニケーション研究会

    2010.4 - 2016.3  副委員長(会計)

     More details

    Committee type:Academic society 

  • 電子情報通信学会 ヴァーバル・ノンヴァーバル・コミュニケーション研究会

    2008.4  専門委員,運営委員

     More details

    Committee type:Academic society 

display all >>