研究者データベース

研究者情報

マスター

アカウント(マスター)

  • 氏名

    飯塚 博幸(イイヅカ ヒロユキ), イイヅカ ヒロユキ

所属(マスター)

  • 人間知・脳・AI研究教育センター

所属(マスター)

  • 人間知・脳・AI研究教育センター

独自項目

syllabus

  • 2021, 大学院共通授業科目(教育プログラム):人間知・脳・AI教育プログラム, Inter-Graduate School Classes(Educational Program):CHAIN Education Program, 修士課程, 大学院共通科目, ディープラーニング、ニューラルネットワーク、機械学習、人工知能、Python
  • 2021, 大学院共通授業科目(教育プログラム):人間知・脳・AI教育プログラム, Inter-Graduate School Classes(Educational Program):CHAIN Education Program, 修士課程, 大学院共通科目, 人文科学 社会科学 神経科学 人工知能 脳 人間 機械学習 統計学
  • 2021, 自律系工学特論, Autonomous Systems Engineering, 修士課程, 情報科学研究科, 自律性,人工生命,力学系
  • 2021, 自律系工学特論, Autonomous Systems Engineering, 修士課程, 情報科学院, 自律性,人工生命,力学系
  • 2021, Introduction to Artificial Intelligence, Big Data, and Cybersecurity for Graduate Students, Introduction to Artificial Intelligence, Big Data, and Cybersecurity for Graduate Students, 修士課程, 情報科学院, artificial intelligence, big data, cryptography, information retrieval, machine learning
  • 2021, 大学院共通授業科目(教育プログラム):人間知・脳・AI教育プログラム, Inter-Graduate School Classes(Educational Program):CHAIN Education Program, 修士課程, 大学院共通科目, 人文科学、社会科学、神経科学、人工知能、脳、神経科学、人間、機械学習、統計モデリング
  • 2021, 大学院共通授業科目(教育プログラム):人間知・脳・AI教育プログラム大学院共通授業科目(教育プログラム, Inter-Graduate School Classes(Educational Program):CHAIN Education Program, 修士課程, 大学院共通科目, 人文科学、社会科学、神経科学、人工知能、脳、人間、機械学習、統計モデリング
  • 2021, 自律系工学特論, Autonomous Systems Engineering, 博士後期課程, 情報科学研究科, 自律性,人工生命,力学系
  • 2021, 自律系工学特論, Autonomous Systems Engineering, 博士後期課程, 情報科学院, 自律性,人工生命,力学系
  • 2021, Introduction to Artificial Intelligence, Big Data, and Cybersecurity for Graduate Students, Introduction to Artificial Intelligence, Big Data, and Cybersecurity for Graduate Students, 博士後期課程, 情報科学院, artificial intelligence, big data, cryptography, information retrieval, machine learning
  • 2021, 情報学 Ⅰ, Introduction to Informatics I, 学士課程, 全学教育, 情報活用,情報社会,情報科学,プログラミング,データサイエンス
  • 2021, 統計学, Statistics, 学士課程, 全学教育, 平均,分散,標準偏差,確率変数,確率分布,母集団,標本,標本分布,点推定,信頼区間,仮説検定
  • 2021, 情報理工学実験Ⅰ, Experiment in Computer Science and Information Technology I, 学士課程, 工学部, プログラミング、計算機システム、データ解析技法
  • 2021, 人工生命と進化型計算, Artificial Life and Evolutionary Computation, 学士課程, 工学部, 学習,進化,自己組織化,自律性,創発,オートポイエシス,フラクタル,セルオートマトン,ライフゲーム,ボイド,仮想生物

researchmap

プロフィール情報

学位

  • 博士(学術)(東京大学)

プロフィール情報

  • プロフィール

    https://sites.google.com/view/hiroyukiiizuka
  • 飯塚, イイヅカ
  • 博幸, ヒロユキ
  • ID各種

    201401015924395996

対象リソース

業績リスト

研究キーワード

  • くすぐり   運動主体感   エージェンシー   感覚統合   意図推定   ヒューマンインターフェイス   バーチャルリアリティ   行動分節化   脳・神経   

研究分野

  • 情報通信 / ヒューマンインタフェース、インタラクション
  • 情報通信 / 知能ロボティクス
  • ライフサイエンス / 基盤脳科学
  • 人文・社会 / 認知科学
  • 情報通信 / データベース

経歴

  • 2023年04月 - 現在 北海道大学 人間知・脳・AI研究教育センター 准教授
  • 2013年10月 - 2023年03月 北海道大学大学院 情報科学研究科 准教授
  • 2008年10月 - 2013年10月 大阪大学 大学院・情報科学研究科 助教
  • 2008年05月 - 2008年09月 大阪大学大学院 情報科学研究科 特任助教
  • 2005年04月 - 2008年03月 日本学術振興会 特別研究員(PD)

学歴

  • 2001年04月 - 2004年03月   東京大学大学院   総合文化研究科   広域科学専攻
  • 1999年04月 - 2001年03月   北海道大学大学院   工学研究科   システム情報工学専攻
  • 1995年04月 - 1999年03月   北海道大学   工学部   情報工学科

受賞

  • 2019年12月 IEEE ALIFE 2019 Best Paper Award
     Analysis of Time Series Generated By Long Short-Term Memory Trained with Adversarial Imitation Learning 
    受賞者: Seiya Yamazaki;Hiroyuki Iizuka;Masahito Yamamoto
  • 2018年12月 The 22nd Asia Pacific Symposium on Intelligent and Evolutionary Systems Best Paper Award
     Combined Method of YOLO and IOU Tracker for Tracking and Identifying Chimpanzee 
    受賞者: Yuichiro Ikeda;Hiroyuki Iizuka;Masahito Yamamoto
  • 2016年09月 日本バーチャルリアリティ学会 論文賞
     往復電流刺激が及ぼす前庭電気刺激の身体動揺増大効果のモデル化 
    受賞者: 青山一真;飯塚博幸;櫻井悟;古川正紘;前田太郎;安藤英由樹
  • 2014年09月 日本バーチャルリアリティ学会 論文賞
     腹腔鏡下手術トレーニングにおける視野共有手法による学習効果とその実証 
    受賞者: 近藤大祐;飯塚博幸;安藤英由樹;小濱和貴;坂井義治;前田太郎

論文

  • A Deep Learning Approach to Understanding the Integration of Tools into Body Schema
    Haruna Kasashima, Wataru Noguchi, Yasumasa Tamura, Masahito Yamamoto, Hiroyuki Iizuka
    ALIFE 2024: Proceedings of the 2024 Artificial Life Conference 2024年07月 [査読有り]
  • Analyzing the Sensibility of Visual Language Models Using an Evolving Image Generation System: Focusing on Color Impressions and Sound Symbolism
    Ryoma Shinto, Hiroyuki Iizuka
    ALIFE 2024: Proceedings of the 2024 Artificial Life Conference 2024年07月 [査読有り]
  • 動物園のゾウを対象にした単一監視カメラ映像によるトラッキング手法の検討と実践
    西岡拳, 野口渉, 飯塚博幸, 山本雅人
    情報処理学会論文誌デジタルプラクティス(TDP) 5 3 60 - 69 2024年07月 [査読有り]
  • 共有モジュール構造をもつRNNによる視覚運動経験を通した階層的空間認知の獲得
    野口渉, 飯塚博幸
    第38回人工知能学会全国大会予稿集 4S1-OS-30a-02 2024年05月 [査読無し][通常論文]
  • Multi-stream Single Network: Efficient Compressed Video Action Recognition with a Single Multi-input Multi-output Network
    Hayto Terao, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    IEEE ACCESS 12 20983 - 20997 2024年02月08日 [査読有り]
  • Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    2023 IEEE International Conference on Development and Learning (ICDL) 2023年11月09日
  • Detecting Eating Behavior of Elephants in a Zoo Using Temporal Action Localization
    Ken Nishioka, Wataru Noguchi, Hiroyuki Izuka, Masahito Yamamoto
    Sensors and Materials 35 11(4) 3927 - 3945 2023年10月 [査読有り][通常論文]
  • Manuel Baltieri, Hiroyuki Iizuka, Olaf Witkowski, Lana Sinapayen, Keisuke Suzuki
    WIREs Cognitive Science 14 6 e1662  2023年07月05日 [査読有り]
     
    Abstract Artificial life is a research field studying what processes and properties define life, based on a multidisciplinary approach spanning the physical, natural, and computational sciences. Artificial life aims to foster a comprehensive study of life beyond “life as we know it” and toward “life as it could be,” with theoretical, synthetic, and empirical models of the fundamental properties of living systems. While still a relatively young field, artificial life has flourished as an environment for researchers with different backgrounds, welcoming ideas, and contributions from a wide range of subjects. Hybrid Life brings our attention to some of the most recent developments within the artificial life community, rooted in more traditional artificial life studies but looking at new challenges emerging from interactions with other fields. Hybrid Life aims to cover studies that can lead to an understanding, from first principles, of what systems are and how biological and artificial systems can interact and integrate to form new kinds of hybrid (living) systems, individuals, and societies. To do so, it focuses on three complementary perspectives: theories of systems and agents, hybrid augmentation, and hybrid interaction. Theories of systems and agents are used to define systems, how they differ (e.g., biological or artificial, autonomous, or nonautonomous), and how multiple systems relate in order to form new hybrid systems. Hybrid augmentation focuses on implementations of systems so tightly connected that they act as a single, integrated one. Hybrid interaction is centered around interactions within a heterogeneous group of distinct living and nonliving systems. After discussing some of the major sources of inspiration for these themes, we will focus on an overview of the works that appeared in Hybrid Life special sessions, hosted by the annual Artificial Life Conference between 2018 and 2022. This article is categorized under:Neuroscience > Cognition Philosophy > Artificial Intelligence Computer Science and Robotics > Robotics
  • Multimodal Plastic Body and Peripersonal Space Representation Developed Through Learning of Visuo-Tactile-Proprioceptive Sensations
    Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    ALIFE 2023: Ghost in the Machine: Proceedings of the 2023 Artificial Life Conference 2023年07月
  • Proprioceptive Drift Can Be Caused by Simple Sensory Prediction
    Harada, K, Noguchi, W, Iizuka, H, Yamamoto, M
    ALIFE 2023: Ghost in the Machine: Proceedings of the 2023 Artificial Life Conference 2023年07月
  • Multiple Timescales RNNを用いた敵対的模倣学習による時系列の複雑化と構造化
    佐々木 太樹, 野口 渉, 飯塚 博幸, 山本 雅人
    情報処理学会第85回全国大会講演論文集 1 349 - 350 2023年03月 [査読無し]
  • 対戦相手の手札推定とカードの適応的優先順位にもとづく UNO の戦術決定
    佐藤 辰哉, 野口 渉, 飯塚 博幸, 山本 雅人
    第22回複雑系マイクロシンポジウム講演論文集 42 - 45 2023年03月 [査読無し]
  • 共有モジュールを用いた予測ネットワークによる対面する他者の視点獲得
    岩崎 智也, 野口 渉, 飯塚 博幸, 山本 雅人
    38 - 41 2023年03月 [査読無し]
  • 共有モジュールを用いた自己他者重ね合わせによる姿勢模倣
    髙橋 光, 野口 渉, 飯塚 博幸, 山本 雅人
    第22回複雑系マイクロシンポジウム講演論文集 29 - 33 2023年03月 [査読無し]
  • 多感覚予測モデルを用いたゴム手錯覚シミュレーションにおける固有感覚ドリフト
    原田 航平, 野口 渉, 飯塚 博幸, 山本 雅人
    第22回複雑系マイクロシンポジウム講演論文集 1 - 4 2023年03月 [査読無し]
  • Junpei Horie, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Artificial Life and Robotics 28 1 148 - 157 2023年02月 [査読有り]
  • Hiroko X. Kondo, Hiroyuki Iizuka, Gen Masumoto, Yuichi Kabaya, Yusuke Kanematsu, Yu Takano
    Biomolecules 13 1 137 - 137 2023年01月09日 [査読有り]
     
    Structure–function relationships in proteins have been one of the crucial scientific topics in recent research. Heme proteins have diverse and pivotal biological functions. Therefore, clarifying their structure–function correlation is significant to understand their functional mechanism and is informative for various fields of science. In this study, we constructed convolutional neural network models for predicting protein functions from the tertiary structures of heme-binding sites (active sites) of heme proteins to examine the structure–function correlation. As a result, we succeeded in the classification of oxygen-binding protein (OB), oxidoreductase (OR), proteins with both functions (OB–OR), and electron transport protein (ET) with high accuracy. Although the misclassification rate for OR and ET was high, the rates between OB and ET and between OB and OR were almost zero, indicating that the prediction model works well between protein groups with quite different functions. However, predicting the function of proteins modified with amino acid mutation(s) remains a challenge. Our findings indicate a structure–function correlation in the active site of heme proteins. This study is expected to be applied to the prediction of more detailed protein functions such as catalytic reactions.
  • Generation of structured spatial patterns through adversarial imitation learning
    Hiroyuki Iizuka, Wataru Noguchi, Taiki Sasaki, Masahito Yamamoto
    The Evolution of Language Proceedings of the Joint Conference on Language Evolution 2022 317 - 324 2022年09月 [査読有り]
  • Hayato Terao, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Machine Learning with Applications 9 100336 100336 - 100336 2022年09月 [査読有り][通常論文]
  • Hiroko X. Kondo, Hiroyuki Iizuka, Gen Masumoto, Yuichi Kabaya, Yusuke Kanematsu, Yu Takano
    Biomolecules 12 9 1172 - 1172 2022年08月24日 [査読有り]
     
    Heme proteins serve diverse and pivotal biological functions. Therefore, clarifying the mechanisms of these diverse functions of heme is a crucial scientific topic. Distortion of heme porphyrin is one of the key factors regulating the chemical properties of heme. Here, we constructed convolutional neural network models for predicting heme distortion from the tertiary structure of the heme-binding pocket to examine their correlation. For saddling, ruffling, doming, and waving distortions, the experimental structure and predicted values were closely correlated. Furthermore, we assessed the correlation between the cavity shape and molecular structure of heme and demonstrated that hemes in protein pockets with similar structures exhibit near-identical structures, indicating the regulation of heme distortion through the protein environment. These findings indicate that the tertiary structure of the heme-binding pocket is one of the factors regulating the distortion of heme porphyrin, thereby controlling the chemical properties of heme relevant to the protein function; this implies a structure–function correlation in heme proteins.
  • Shuto Kuriyama, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the Artificial Life Conference 2022 407 - 409 2022年07月 [査読有り][通常論文]
  • Hiroyuki Iizuka, Taiki Sasaki, Wataru Noguchi, Masahito Yamamoto
    Proceedings of the Artificial Life Conference 2022 298 - 300 2022年07月 [査読有り][通常論文]
  • Shimada Yasuhiro, Noguchi Wataru, Iizuka Hiroyuki, Yamamoto Masahito
    Proceedings of the Artificial Life Conference 2022 147 - 154 2022年07月 [査読有り][通常論文]
  • 野口 渉, 飯塚 博幸, 山本 雅人
    システム/制御/情報 66 4 139 - 144 2022年04月 [査読無し][招待有り]
  • 畳み込みニューラルネットワークを用いた色の恒常性の学習によるドレスの色の錯視の再現
    市川貴大, 野口渉, 飯塚博幸, 山本雅人
    情報処理学会第84回全国大会 2 861 - 862 2022年03月 [査読無し]
  • Superposition mechanism as a neural basis for understanding others
    Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto, Shigeru Taguchi
    Scientific Reports 12 2859 2022年02月 [査読有り]
  • Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Advanced Robotics 36 1-2 85 - 99 2022年01月02日 [査読有り]
  • The shared-module neural network for self/other differentiation and symbol grounding
    Hiroyuki Iizuka
    Proceedings of the twenty-seventh international symposium on Artificial life and Robotics 1124 1125 2022年01月 [査読有り]
  • Increasing complexity of symbol sequences in adversarial imitation learning experiment
    Kazufumi Nomura, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the twenty-seventh international symposium on Artificial life and Robotics 231 - 236 2022年01月 [査読有り]
  • Understanding intention of others through visual predictive learning based on superposition mechanism
    Katsuki Auchi, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the twenty-seventh international symposium on Artificial life and Robotics 1812 - 1817 2022年01月 [査読有り]
  • 教師ありBarlow Twinsによる少データ手話単語認識のための事前学習
    荒木関 渉, 野口 渉, 飯塚 博幸, 山本 雅人
    第22回公益社団法人計測自動制御学会システムインテグレーション部門講演会 629 - 634 2021年12月 [査読無し]
  • I3Dを用いたツシマヤマネコの発情兆候検出
    韓靖威, 足立樹, 小泉那於, 野口渉, 飯塚博幸, 山本雅人
    ロボティクス・メカトロニクス講演会2021 (ROBOMECH2021)講演論文集 2021年05月 [査読無し]
  • カメラ画像を用いた深層学習による水族館の水透明度判定
    高一, 佐藤聡, 西村究, 野口渉, 飯塚博幸, 山本雅人
    ロボティクス・メカトロニクス講演会2021 (ROBOMECH2021)講演論文集 2021年05月 [査読無し]
  • Kazuma Aoyama, Nobuhisa Miyamoto, Satoru Sakurai, Hiroyuki Iizuka, Makoto Mizukami, Masahiro Furukawa, Taro Maeda, Hideyuki Ando
    IEEE Access 9 106714 - 106724 2021年 [査読有り]
  • Complexity Growth by Cooperative Interaction in Adversarial Learning
    Taiki Sasaki, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the twenty-sixth international symposium on Artificial life and Robotics 78 - 83 2021年01月 [査読有り][通常論文]
  • Multi-agent formation by learning communication signals using DIAL in a distributed multi-agent environment
    Hironobu Horiuchi, Wataru Noguchi, Iizuka Hiroyuki, Masahito Yamamoto
    Proceedings of the twenty-sixth international symposium on Artificial life and Robotics 308 - 312 2021年01月 [査読有り][通常論文]
  • 周辺地域の予測レーダ雨量を用いた深層学習による水路の水位予測
    石川拓実, 阿曽克司, 羽黒厚志, 野口渉, 飯塚博幸, 山本雅人
    第21回計測自動制御学会システムインテグレーション部門講演会 2080 - 2083 2020年12月 [査読無し][通常論文]
  • カメラ映像を用いたYOLOとI3Dによる飼育動物の行動分類
    加藤圭吾, 野口渉, 飯塚博幸, 山本雅人
    "第21回計測自動制御学会システムインテグレーション部門講演会 " 2084 - 2087 2020年12月 [査読無し][通常論文]
  • Hayato Terao, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    2020 International Conference on Vision, Image and Signal Processing(ICVISP 2020) 14 - 6 2020年12月 [査読有り][通常論文]
  • DIALを用いた通信シグナルの学習によるマルチエージェントフォーメーション
    堀内宏信, 野口渉, 飯塚博幸, 山本雅人
    19回情報科学技術フォーラム(FIT2020)講演論文集, 第2分冊 265 - 266 2020年09月 [査読無し][通常論文]
  • 複数の代表点を用いたSDMMに基づく半教師あり行動認識手法
    寺尾 颯人, 野口 渉, 飯塚 博幸, 山本 雅人
    19回情報科学技術フォーラム(FIT2020)講演論文集, 第3分冊 165 - 166 2020年09月 [査読無し][通常論文]
  • 深層学習モデルによる自己と他者の共有身体イメージの獲得
    野口 渉, 飯塚 博幸, 山本 雅人
    19回情報科学技術フォーラム(FIT2020)講演論文集, 第2分冊 299 - 300 2020年09月 [査読無し][通常論文]
  • Olaf Witkowski, Takashi Ikegami, Nathaniel Virgo, Mizuki Oka, Hiroyuki Iizuka
    ARTIFICIAL LIFE 26 1 1 - 4 2020年04月
  • Learning of Evaluation Function in Digital Curling Considering the Probability of Scores at Each End
    Kotaro Ataka, Wataru Noguchi, Iizuka Hiroyuki, Masahito Yamamoto
    AAAI20 Workshop, Artificial Intelligence in Team Sports 1 - 6 2020年02月 [査読有り][通常論文]
  • MixMatch based semi-supervised learning for video action recognition using 3D convolutional neural networks
    Hayato Terao, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the twenty-fifth international symposium on Artificial life and Robotics 305 - 310 2020年01月 [査読有り][通常論文]
  • Efffect of Reactivity Artificial Medaka on Real Medaka
    Katsuki Auchi, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the twenty-fifth international symposium on Artificial life and Robotics 86 - 89 2020年01月 [査読有り][通常論文]
  • Kenji Hirata, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 2020 IEEE/SICE International Symposium on System Integration 201 - 204 2020年01月 [査読有り][通常論文]
  • Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 2020 IEEE/SICE International Symposium on System Integration 182 - 187 2020年01月 [査読有り][通常論文]
  • Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    Artificial Life and Robotics 25 1 124 - 132 2020年01月 [査読有り][通常論文]
  • Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 2019 IEEE Symposium Series on Computational Intelligence (IEEE SSCI 2019) 224 - 228 2019年12月 [査読有り][通常論文]
  • 分割した集団の学習による役割分担の獲得
    荒木関 渉, 野口 渉, 飯塚 博幸, 山本 雅人
    情報処理北海道シンポジウム2019 2019年10月 [査読無し][通常論文]
  • Hiroyuki Iizuka, Yosuke Nakamoto, Masahito Yamamoto
    Proceedings of the Artificial Life Conference 2019 50 - 51 2019年07月 [査読有り][通常論文]
  • Wataru Noguchi, Hiroyuki Iizuka, Shigeru Taguchi, Masahito Yamamoto
    Proceedings of the Artificial Life Conference 2019 531 - 532 2019年07月 [査読有り][通常論文]
  • 3次元畳み込みニューラルネットワークを用いた店舗における顧客行動の分類
    貝塚 裕, 飯塚 博幸, 山本 雅人
    ロボティクス・メカトロニクス講演会2019 (ROBOMECH2019) 1A1-M05  2019年06月 [査読無し][通常論文]
  • UAVの空撮画像を利用した産業廃棄物最終処分場の容量推定
    賀川祐太郎, 飯塚博幸, 山本雅人
    第51回計測自動制御学会北海道支部学術講演会論文集 27 - 30 2019年03月 [査読無し][通常論文]
  • 不確定ゲームにおける深層強化学習による評価関数の獲得
    安宅耕太郎, 飯塚博幸, 山本雅人
    第18回複雑系マイクロシンポジウム(CSMS2019) 62 - 65 2019年03月 [査読無し][通常論文]
  • LSTM Autoencoderを用いた半教師あり学習による行動分類のための特徴ベクトル抽出手法
    寺尾颯人, 飯塚博幸, 山本雅人
    第18回複雑系マイクロシンポジウム(CSMS2019) 25 - 30 2019年03月 [査読無し][通常論文]
  • Deep Q-Networkを用いたコミュニケーションシグナルの分化
    堀内宏信, 飯塚博幸, 山本雅人
    第18回複雑系マイクロシンポジウム(CSMS2019) 76 - 80 2019年03月 [査読無し][通常論文]
  • Naohiro Shibuya, Hiroyuki Iizuka, Masahito Yamamoto
    2019 IEEE/SICE International Symposium on System Integration (SII2019) 764 - 769 2019年01月 [査読有り][通常論文]
  • Establishing Mutual Interaction between Artificial and Real Medaka
    Yosuke Nakamoto, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the Twenty-Fourth International Symposium on Artificial Life and Robotics (AROB2019) 633 - 637 2019年01月 [査読有り][通常論文]
  • Adversarial Imitation Learning of Bird Song Modeled with Recurrent Neural Network
    Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the Twenty-Fourth International Symposium on Artificial Life and Robotics (AROB2019) 162 - 167 2019年01月 [査読有り][通常論文]
  • Modeling place and head-direction cells self-organized in different environments using recurrent neural networks
    Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the Twenty-Fourth International Symposium on Artificial Life and Robotics (AROB2019) 67 - 72 2019年01月 [査読有り][通常論文]
  • Mutual learning between human and machine to develop a human-machine interface to operate multi-DOF robots
    Kei Takada, Hiroyuki Iizuka, Subhransu Maji, Masahito Yamamoto
    Proceedings of the Twenty-Fourth International Symposium on Artificial Life and Robotics (AROB2019) 397 - 403 2019年01月 [査読有り][通常論文]
  • Kei Takada, Hiroyuki Iizuka, Masahito Yamamoto
    IEEE transactions on Games 12 1 63 - 73 2019年01月 [査読有り][通常論文]
  • Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Advanced Robotics 33 11 539 - 549 2019年01月 [査読有り][通常論文]
  • Hiroya Nagata, Soichiro Yokoyama, Tomohisa Yamashita, Hiroyuki Iizuka, Masahito Yamamoto, Keiji Suzuki, Hidenori Kawamura
    Journal of Robotics and Mechatronics 30 3 390 - 396 2018年06月 
    Proportional-integral-derivative (PID) controllers are a classical control algorithm that are still widely used owing to their simplicity and accuracy. However, tuning the three parameters is difficult. No methods have been known to determine the exact ideal combination of the P, I, and D gains. Moreover, controlling a system that contains dynamics changes over time using fixed parameters is difficult. A self-tuning neuro-PID controller is applied to a balloon robot for indoor entertainment to enhance its accuracy in following a target trajectory. Our experiment shows the effectiveness of the neuro-PID controller over conventional hand-tuned PID controller.
  • Hiroya Nagata, Soichiro Yokoyama, Tomohisa Yamashita, Hiroyuki Iizuka, Masahito Yamamoto, Keiji Suzuki, Hidenori Kawamura
    JOURNAL OF ROBOTICS AND MECHATRONICS 30 3 390 - 396 2018年06月 
    Proportional-integral-derivative (PID) controllers are a classical control algorithm that are still widely used owing to their simplicity and accuracy. However, tuning the three parameters is difficult. No methods have been known to determine the exact ideal combination of the P, I, and D gains. Moreover, controlling a system that contains dynamics changes over time using fixed parameters is difficult. A self-tuning neuro-PID controller is applied to a balloon robot for indoor entertainment to enhance its accuracy in following a target trajectory. Our experiment shows the effectiveness of the neuro-PID controller over conventional hand-tuned PID controller.
  • Kei Takada, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings - 2017 Conference on Technologies and Applications of Artificial Intelligence, TAAI 2017 196 - 201 2018年05月09日 [査読有り][通常論文]
     
    An evaluation function in the board game decides the next move for computer AIs, and a high accurate evaluation function leads to a strong computer AI. Recently, the evaluation function using convolutional neural network(CNN) by supervised learning shows high evaluation accuracy. Supervised learning cannot exceed the teachers, but there must be a possibility to create a more accurate evaluation function by using reinforcement learning. In this paper, we proposed an evaluation function using CNN and reinforcement learning with games of self-play in Hex. The proposed evaluation function is tested with the previous evaluation function and world-champion program MoHex2.0. The results show that evaluation accuracy of the proposed evaluation function is higher than the previous evaluation functions, and proposed computer Hex algorithm EZO-CNN obtained a win rate of 60.0% against MoHex2.0 even though the search time of EZO-CNN is shorter than MoHex2.0.
  • Takashi Ikegami, Nathaniel Virgo, Olaf Witkowski, Mizuki Oka, Reiji Suzuki, Hiroyuki Iizuka
    2018 CONFERENCE ON ARTIFICIAL LIFE (ALIFE 2018) 1 - 4 2018年
  • Hiroya Nagata, Soichiro Yokoyama, Tomohisa Yamashita, Hiroyuki Iizuka, Masahito Yamamoto, Keiji Suzuki, Hidenori Kawamura
    J. Robotics Mechatronics 30 3 390 - 396 2018年
  • Kei Takada, Hiroyuki Iizuka, Masahito Yamamoto
    Communications in Computer and Information Science 818 19 - 33 2018年 [査読有り][通常論文]
     
    In recent years, a move evaluation model using a convolutional neural network (CNN) has been proposed for Go, and it has been shown that CNN can learn professional human moves. Hex is a two-player connection game, which is included in the Computer Olympiad. It is important to consider cell adjacency on the board for a better Hex strategy. To evaluate cell adjacency from various perspectives properly, we propose a CNN model that evaluates all candidate moves by taking as input all sets consisting of 3 mutually adjacent cells. The proposed CNN model is tested against an existing CNN model called “NeuroHex,” and the comparison results show that our CNN model is superior to NeuroHex on a 13×13 board even though our CNN model is trained on an 11×11 board. We also use the proposed model as an ordering function and test it against the world-champion Hex program “MoHex 2.0” on an 11×11 board. The results show that the proposed model can be used as a better ordering function than the ordering function created by minimax tree optimization, and we obtained a win rate of 49.0% against MoHex 2.0 (30 s/move).
  • Combined Method of YOLO and IOU Tracker for Tracking and Identifying Chimpanzee
    Yuichiro Ikeda, Hiroyuki Iizuka, Masahito Yamamoto
    The 22nd Asia Pacific Symposium on Intelligent and Evolutionary Systems (IES 2018) 110 - 115 2018年 [査読有り][通常論文]
  • 手話動作分類におけるRCNNモデルの性能評価と内部状態解析
    松田 啓佑, 飯塚 博幸, 山本雅人
    2018年度 人工知能学会全国大会 2631 - 2634 2018年 [査読無し][通常論文]
  • 畳み込みニューラルネットワークによるチンパンジーの個体識別
    池田 宥一郎, 飯塚 博幸, 山本雅人
    2018年度 人工知能学会全国大会(第32回) 93 - 95 2018年 [査読無し][通常論文]
  • Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 2018 Conference on Artificial Life 659 - 664 2018年 [査読有り][通常論文]
     
    Bird song is one of the phenomena that increase in complexity through evolution. A complex song is known to be advantageous for survivability and birds are known to learn how to sing a song from each other. From these facts, we have a hypothesis that adversarial imitation learning plays a major role in the evolution process of a complex song. There is a previous study that demonstrates the complexation of a bird song time series by modeling the process of adversarial imitation learning using a logistic map. However, the real bird songs have much variety and time dependencies, like grammar. Therefore, in this study, adversarial imitation learning is modeled using an artificial neural network that can approximate any function. The network learns adversarial imitation using the gradient descent method. By making such changes, the results of our study show that the generated bird songs evolve through the process of adversarial imitation learning to chaos, as seen in the previous models.
  • Hiroyuki Iizuka, Yosuke Namamoto, Masahito Yamamoto
    Proceedings of the 2018 Conference on Artificial Life 179 - 185 2018年 [査読有り][通常論文]
     
    Swarms of birds and fish produce well-organized behaviors even though each individual only interacts with their neighbors. Previous studies attempted to derive individual interaction rules using heuristic assumptions from data on captured animals. We propose a machine learning method to obtain the sensorimotor mapping mechanism of individuals directly from captured data. Data on swarm behaviors in fish was captured, and individual positions are determined. The sensory inputs and motor outputs are estimated and used as training data. A simple feedforward neural network is trained to learn the sensorimotor mapping of individuals. The trained network is implemented in the simulated environment and resulting swarm behaviors are investigated. As a result, our trained neural network could reproduce the swarm behavior better than the Boids model. The reproduced swarm behaviors are evaluated in terms of three different measures, and the difference from the Boids model is discussed.
  • Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 2018 Conference on Artificial Life 147 - 154 2018年 [査読有り][通常論文]
     
    Animals develop spatial recognition through visuomotor integrated experiences. In nature, animals change their behavior during development and develop spatial recognition. The developmental process of spatial recognition has been previously studied. However, it is unclear how behavior during development affects the development of spatial recognition. To investigate the effect of movement pattern (behavior) on spatial recognition, we simulated the development of spatial recognition using controlled behaviors. Hierarchical recurrent neural networks (HRNNs) with multiple time scales were trained to predict visuomotor sequences of a simulated mobile agent. The spatial recognition developed with HRNNs was compared for various values of randomness of the agent's movement. The experimental results show that spatial recognition was not developed for movements with a randomness that was too small or too large but for movements with intermediate randomness.
  • Deep Reinforcement Learning Method Enabling Stable Learning in Multi-Agent Environment
    Kenji Hirata, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 22nd Asia Pacific Symposium on Intelligent and Evolutionary Systems (IES 2018) 163 - 168 2018年 [査読有り][通常論文]
  • Differentiation of communication signals to establish cooperation using Deep Q-Network
    Hironobu Horiuchi, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of the 22nd Asia Pacific Symposium on Intelligent and Evolutionary Systems (IES 2018) 156 - 162 2018年 [査読有り][通常論文]
  • Hiroya Nagata, Stefan Walke, Soichiro Yokoyama, Tomohisa Yamashita, Hiroyuki Iizuka, Masahito Yamamoto, Keiji Suzuki, Hidenori Kawamura
    Artificial Life and Robotics 23 2 192 - 199 2017年12月 [査読有り][通常論文]
  • Naohiro Shibuya, Hiroyuki Iizuka, Masahito Yamamoto
    Artificial Life and Robotics 23 2 225 - 240 2017年12月 [査読有り][通常論文]
  • Hiroki Kojima, Tom Froese, Mizuki Oka, Hiroyuki Iizuka, Takashi Ikegami
    FRONTIERS IN PSYCHOLOGY 8 1778 2017年10月 [査読有り][通常論文]
     
    It is not yet well understood how we become conscious of the presence of other people as being other subjects in their own right. Developmental and phenomenological approaches are converging on a relational hypothesis: my perception of a "you" is primarily constituted by another subject's attention being directed toward "me." This is particularly the case when my body is being physically explored in an intentional manner. We set out to characterize the sensorimotor signature of the transition to being aware of the other by re-analyzing time series of embodied interactions between pairs of adults (recorded during a "perceptual crossing" experiment). Measures of turn-taking and movement synchrony were used to quantify social coordination, and transfer entropy was used to quantify direction of influence. We found that the transition leading to one's conscious perception of the other's presence was indeed characterized by a significant increase in one's passive reception of the other's tactile stimulations. Unexpectedly, one's clear experience of such passive touch was consistently followed by a switch to active touching of the other, while the other correspondingly became more passive, which suggests that this intersubjective experience was reciprocally co-regulated by both participants.
  • Ming Chang, Hiroyuki Iizuka, Hideki Kashioka, Yasushi Naruse, Masahiro Furukawa, Hideyuki Ando, Taro Maeda
    PLOS ONE 12 6 e0178694  2017年06月 [査読有り][通常論文]
     
    When people learn foreign languages, they find it difficult to perceive speech sounds that are nonexistent in their native language, and extensive training is consequently necessary. Our previous studies have shown that by using neurofeedback based on the mismatch negativity event-related brain potential, participants could unconsciously achieve learning in the auditory discrimination of pure tones that could not be consciously discriminated without the neurofeedback. Here, we examined whether mismatch negativity neurofeedback is effective for helping someone to perceive new speech sounds in foreign language learning. We developed a task for training native Japanese speakers to discriminate between 'l' and 'r' sounds in English, as they usually cannot discriminate between these two sounds. Without participants attending to auditory stimuli or being aware of the nature of the experiment, neurofeedback training helped them to achieve significant improvement in unconscious auditory discrimination and recognition of the target words 'light' and 'right'. There was also improvement in the recognition of other words containing 'l' and 'r' (e.g., 'blight' and 'bright'), even though these words had not been presented during training. This method could be used to facilitate foreign language learning and can be extended to other fields of auditory and clinical research and even other senses.
  • Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    ADAPTIVE BEHAVIOR 25 3 129 - 146 2017年06月 [査読有り][通常論文]
     
    Animals develop and use cognitive maps, which are internal models of the external environment, to understand the spatial characteristics of their natural environment. Previous studies have shown that a hierarchical structure of recurrent neural networks contributes to the extraction of high-level concepts in sequential sensorimotor experiences. However, the previous studies did not focus on the spatial aspects of these experiences and did not acquire cognitive maps. We modified previous models and trained the proposed model with the visuomotor experiences of an agent in a simulated two-dimensional environment. The proposed model was trained to predict future visual and motion inputs even when only one modality was provided (crossmodal prediction). The proposed model correctly predicted visual images, even when the agent experienced unknown paths. Comparisons of the crossmodal predictions of the models under different conditions revealed that the crossmodal predictions related to motion resulted in self-organization of the cognitive map. Further experiments of mental simulation abilities showed that two-way crossmodal predictions (from vision and motion only) were required for consistent generation of vision and motion. These results indicated that predictive learning involving integrated vision and motion was necessary for self-organization of spatial recognition with a cognitive map.
  • Zhuoran Xu, Hiroyuki Iizuka, Masahito Yamamoto
    SOFT COMPUTING 21 5 1327 - 1345 2017年03月 [査読有り][通常論文]
     
    Many real-world problems are multimodal, which means algorithms should have the ability to find all or most of the multiple solutions as opposed to a single best solution. Niching is the technique that can help evolutionary algorithms to find multiple solutions. Attraction basin sphere estimation (ABSE) is a newly proposed niching method which has the power of inferring the shape of fitness landscapes by spending some extra evaluations. However, when we apply ABSE to genetic algorithms, those extra evaluations lead to an efficiency problem. However, we notice that the combination of ABSE and covariance matrix adaptation evolution strategy will not cause the efficiency problem and have a good performance. This paper implements this idea and performs experiments on a benchmark set provided by CEC 2013 niching methods competition. The algorithm is compared with the best 5 algorithms in the competition. The results show that the proposed algorithm obtains a good result. The features of the proposed algorithm are also discussed in detail.
  • 西村 亮, 飯塚 博幸, 山本 雅人
    精密工学会学術講演会講演論文集 2017 27 - 28 公益社団法人 精密工学会 2017年 
    DCGANは画像生成モデルの学習手法として注目を集めており,画像の生成には潜在空間上のベクトルを用いる.潜在空間上には学習した画像の特徴が連続的に分布しており,潜在空間上の2点の中間点では2つの画像の特徴を併せもつような画像となる.本研究では既知の画像を潜在空間へ局在配置させる学習を行い,潜在空間と生成画像を関係づけることで異なる性質の画像の融合を試みる.
  • 小川 純, 飯塚 博幸, 山本 雅人
    精密工学会学術講演会講演論文集 2017 15 - 16 公益社団法人 精密工学会 2017年 
    異なる環境における物体追跡を目的とするソフトロボットの形状最適化を行う.ニューラルネットワークによりボクセルの生成及び物性を求め,ロボットを設計する.シミュレーションを用いて地上及び水中における物体追跡距離を最大化するロボット形状を学習する.異なる最適化条件を検証した結果,単一環境のみで最適化されたロボットより両環境におけるパレート最適化により得られたロボットの方が高い評価値を持つことを示した.
  • 杉中 出帆, 飯塚 博幸, 山本 雅人
    ロボティクス・メカトロニクス講演会講演概要集 2017 2P2 - F02 一般社団法人 日本機械学会 2017年 

    Many robots are pervading environments of human daily life. It is crucial for those robots to estimate the current self-positions. This is called robot localization. In this paper, we propose a method using a recurrent convolutional neural network (RCNN), which is known as one of deep learning, to achieve robot localization. RCNN is a neural network model that has a convolutional architecture known as CNN with recurrent nodes. We train RCNN to estimate the current position of a robot from the view images of the first person perspectives. Our experimental results show that the estimation error decrease when the successive view images are given and it can estimate the current position accurately.

  • Yosuke Nakamoto, Hiroyuki Iizuka, Masahito Yamamoto
    2017 IEEE/SICE INTERNATIONAL SYMPOSIUM ON SYSTEM INTEGRATION (SII) 847 - 852 2017年 
    The swarm of birds and fish produces the well organized behavior even though each individual is only interacting with neighbors. In order to form the swarm, there should be some rules that each individual is following. In this paper, we examine whether the neural network can learn the individual behavioral rules and reproduce the swarm behavior. In the experiment, boids swarm and real fish swarm are used for the target of learning and reproduction. As a result, the neural network could learn and reproduce the behavior almost perfectly for the boids swarm, and partially for the real fish swarm.
  • Izuho Suginaka, Hiroyuki Iizuka, Masahito Yamamoto
    2017 21ST ASIA PACIFIC SYMPOSIUM ON INTELLIGENT AND EVOLUTIONARY SYSTEMS (IES) 95 - 100 2017年 
    Mobile robot localization has been considered to be an important task in the field of robotics research. It is known that it is difficult to estimate the self-position in dynamic environments where the positions of objects used as landmarks change. In this paper, we propose a robust method to estimate self-position from the first person view captured by a camera on a robot using Recurrent Convolutional Neural Networks (RCNN), which is a neural network model that has a convolutional architecture known as CNN with recurrent nodes. The RCNN receives images and directly estimates the positions of the robot. Our proposed method is evaluated in simulated environments. Our experiments show that RCNN model can estimate the self position of the robot with high accuracy even if some objects move to different positions, that is, it has a robustness against objects obstructing visibility.
  • Yosuke Nakamoto, Hiroyuki Iizuka, Masahito Yamamoto
    847 - 852 2017年
  • Sohtaroh Saitoh, Hiroyuki Iizuka, Masahito Yamamoto
    INTELLIGENT AND EVOLUTIONARY SYSTEMS, IES 2016 8 361 - 369 2017年 [査読有り][通常論文]
     
    Perceptual Crossing experiment is the method for analyzing human interaction with low degrees of freedom of perception and motion. In this experiment, participants need to establish a cooperative interaction to achieve a task. In this study, we introduce time delays of participant's perception in this experiment, and evaluate the stability of human social interaction against the time delay. Our result shows that human social interactions have the stability against the time delays by changing their behaviors adaptively. We also investigate if the stability can be enhanced by modulating the participant's behavior conversely. It is shown that it is possible to increase the stability of social interaction by slowing down behaviors of participants.
  • Intuitive control of mobile robots, an architecture for autonomous adaptive dynamic behaviour integration
    Christos Melidis, Hiroyuki Iizuka, Davide Marocco
    Cognitive Processing 1 - 20 2017年 [査読有り][通常論文]
  • Evolution of Swarm Formation Using Homogeneous Multi-Robots
    Naohiro Shibuya, Hiroyuki Iizuka, Masahito Yamamoto
    Proceedings of The 2nd International Symposium on Swarm Behavior and Bio-Inspired Robotics 128 - 133 2017年 [査読有り][通常論文]
  • Ryo Nishimura, Hiroyuki Iizuka, Masahito Yamamoto
    BICT 2015 - 9th EAI International Conference on Bio-Inspired Information and Communications Technologies 16 9 e3  2016年05月24日 [査読有り][通常論文]
     
    Social interaction is the basic ability for animals to survive. It is difficult for a machine to interact with human or other animals because it is not clear how the machine should interact. This paper examines whether an artificial dot controlled by a machine can interact with a medaka and induce a desired behavior. The dot is displayed on a monitor. We use deep Q network (DQN) to learn how to move the dot. As a result, the DQN could learn some basic elements to interact with the medaka and the desired behavior could be induced.
  • 齋藤 健太郎, 飯塚 博幸, 山本 雅人
    精密工学会学術講演会講演論文集 2016 575 - 576 公益社団法人 精密工学会 2016年 
    ロボットの構造が複雑になるにつれて,操縦するためのインタフェースも複雑なものとなり,ユーザは複雑な操縦方法の学習を強いられる.本研究では人間の学習負荷が低い簡易なインタフェースにもかかわらず,複雑な構造のロボットを操縦できるインタフェースの作成手法を提案する.動作の分節化を利用して複数のインタフェースを切り替えて学習を行うことで,さまざまな動作を可能とするインタフェースを構築し評価する.
  • Sohtaroh Saitoh, Hiroyuki Iizuka, Masahito Yamamoto
    361 - 369 2016年
  • Jun Ogawa, Hiroyuki Iizuka, Masahito Yamamoto, Masashi Furukawa
    INTELLIGENT AUTONOMOUS SYSTEMS 13 302 821 - 830 2016年 [査読有り][通常論文]
     
    This paper describes new approaches for classifying twist of seaweeds. There are no evaluation measures of the twist formation of complicated objects quantitatively because the definition of a qualitative twist is a difficult problem. The twist of seaweeds is one of these problems. In this paper, we propose three factors (physical, geometric, and time factor of twist) for characterizing the twist state, and we develop the twist-state classifier based on these factors. Additionally, the analysis experiment verifies how the classifier shows the classification accuracy of twist state.
  • 本庄 将也, 飯塚 博幸, 山本 雅人, 古川 正志
    日本知能情報ファジィ学会誌「知能と情報」 28 4 744 - 755 日本知能情報ファジィ学会 2016年 [査読有り][通常論文]
     
    近年注目されている実数値最適化手法の一つに粒子群最適化(Particle Swarm Optimization, PSO)がある.PSOは群知能の一種であり,複数の探索単位(粒子)が互いに情報共有を行いながら解の探索を行う.多点探索を行うメタヒューリスティクスとしては遺伝的アルゴリズム(Genetic Algorithm, GA)が有名であるが,多くの実数値最適化問題においてPSOのほうがGAに比べて高速に良い解を発見できることが知られている.本研究では,組合せ最適化問題の一種である巡回セールスマン問題(Traveling Salesman Problem, TSP)に対して短時間で良い解を得ることを目的として,PSOを基にしたアルゴリズムである挿入操作PSO戦略を提案する.提案手法では,粒子の解候補は実数値ベクトルではなく巡回路として表現され,粒子間の相互作用は部分経路挿入によって行われる.本論文では,挿入操作PSO戦略について説明し,数値計算実験からパラメータと得られる解の良さと必要な時間の関係について調査し,パラメータ調整の指針を示す.また,各ベンチマーク問題に対して提案手法とGAなどの代表的なメタヒューリスティクスを適用し,提案手法がこれらの手法より短時間で良い解を求められることを示す.
  • 利根川 凛, 飯塚 博幸, 山本 雅人, 古川 正志, 大内 東
    情報処理学会論文誌 57 6 1565 - 1575 2016年 [査読有り][通常論文]
     
    津波の被害に代表される大規模な家屋の倒壊が起こる災害時において,被災状況を把握できずに起きる復旧の遅延が問題となっている.津波により建造物の倒壊が起こり大量のがれきが発生すると,早急に輸送手段や仮置き場の位置や数の手配が必要になる.しかしながら,現状では被災地での被害状況の把握,被災がれき量の推定について短時間で行う手法は確立されていない.本研究では,震災後に撮影される被災地の空撮画像を複数用い,被災地の地図情報から家屋のある範囲を抽出し,畳み込みニューラルネットワークによる学習から,家屋倒壊判別を行う.結果として,東日本大震災で被災地となった10地域の空撮画像を利用し,約86%の正答率を得た.また,使用地域の中で,テストデータの判別正答率の悪かった地域のみをテストデータに絞った場合でも判別を行い,前実験の結果と比較して,より多くの地域,および,条件の画像データを用意すれば,判別器の正答率が向上する可能性を示した.When a massive earthquake that causes Tsunami happens, it is too difficult to get a full picture of the damages caused by the disaster and it causes the delay to recover from it. The debris generated by Tsunami or broken houses block roads and all transportation for people, rescue supply and bringing debris stop. In order to remove the debris and clear the roads, we need to decide which and how many spaces should be allocated for temporal storages immediately. However, there is no method to estimate the amount of generated debris and to get a full picture of the damage for a short time. This paper proposes a novel method that uses a convolutional neural network to classify images taken by a plane into damage or no-damage automatically. The network is trained with the images of 10 different areas stricken by the Great East Japan Earthquake. Our results shows that the trained network can correctly answer with about 86% and we found that the correct rates changes in response to the extents of the damages in different areas. The possibility to create a better network classifier is discussed in the end.
  • Jun Ogawa, Hiroyuki Iizuka, Masahito Yamamoto, Masashi Furukawa
    Journal of Robotics and Mechatronics 28 1 40 - 49 2016年 [査読有り][通常論文]
     
    We discuss novel approaches to the control of seaweed tangle formations in stirrer cultivation. Cultivating seaweed is one important way to avoid such formation. Because defining such formation is difficult based on human recognition alone, there is currently no quantitative evaluation criterion for formation. We develop physical simulation for analyzing formations in a water flow field and model three factors – physical, geometric and time – for characterizing formations. Our criterion is that formations are created by using these factors as input to a nonlinear support vector machine. To show the effectiveness of our simulation and criteria, we confirm the control effects of the water flow in simulation and the real world. Results show that our simulation model is useful for avoiding such formation in the real world.
  • Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    EAI Endorsed Transactions on Security and Safety 16 10 e1 - 1 2016年 [査読有り][通常論文]
  • 不確定性を含むデジタルカーリングにおけるゲーム木探索
    加藤修, 飯塚博幸, 山本雅人
    情報処理学会論文誌ジャーナル 57 11 2354 - 2364 2016年 [査読有り][通常論文]
  • 南 賢一, 飯塚 博幸, 古川 正志, 山本 雅人
    人工知能学会論文誌 30 6 737 - 744 2015年10月27日 [査読有り][通常論文]
     
    The distribution center became more important because the growth of the internet enabled us to buy various products easily on the Internet. In the distribution center, one of the main works is order picking which is to carry the products from shelves of the warehouse according to the orders received from customers. Efficient order picking reduces the large amount of work and the improvement of the order picking is necessary in a larger distribution center. It is important for order picking not only to give workers a good tour plan but also to assign products to the storage shelves in the distribution center. It is known that it is effective to consider frequency order and frequency of co-occurrence at the same time for storage location assignment. The conventional studies proposes the class-based method for frequency order and evaluation function to evaluate frequency order and co-occurrence of orders at the same time. This study proposes a novel method how to assign products to the storage using a co-occurrence network of ordered products and self-organizing map on network topology of the warehouse. Our proposed method is compared with the conventional studies and our experiments show that our proposed method can be more effective than the conventional methods.
  • 高田 圭, 本庄 将也, 飯塚 博幸, 山本 雅人
    人工知能学会論文誌 30 6 729 - 736 2015年10月27日 [査読有り][通常論文]
     
    The game of Hex is the board game with simple rules and is classified as a two-player, zero-sum, logical perfect information game. The game proceeds by putting their pieces in turn on empty cells of the board. A player wins if the player connects the two opposing sides of the board with their own color pieces. Our previous study clarified that it is effective to develop a computer Hex strategies with the network characteristics as the evaluation function to evaluate the board states from the global and local perspectives and showed that there is the best parameter to decide the ratio between global and local evaluation during a match. In order to go beyond the strategy, we hypothesize that the ratio must change during a match depending on the board states as human players differently evaluate the board states at the beginning, middle and final stages. First, we examine the hypothesis whether the better wining rate can be achieved by changing the ratio of global and local evaluation and propose a novel computer Hex program that can evaluate the board states while changing the global and local evaluation by recognizing the board states with SVM. Our proposed method is evaluated with the current world-champion program called MoHex.
  • Hiroyuki Iizuka, Sohtaroh Saitoh, Davide Marocco, Masahito Yamamoto
    HAI 2015 - Proceedings of the 3rd International Conference on Human-Agent Interaction 217 - 219 2015年10月21日 [査読有り][通常論文]
     
    This paper investigates time-delay effects of the human social interaction to understand how human can adapt to the time delay, which will be required in software agents to establish a harmonic interaction with human. We performed the minimal experiments of social interaction called perceptual crossing experiments with time delay. Our result shows that the social interaction breaks down when the total amount of time delay is given more than about one second. However, the interaction breaks more easily when the time delay is given to both participants than to either participant.
  • Water Flow Control for Inhibiting Seaweed Twist in Real Environment and Physical Simulation
    Jun Ogawa, Hiroyuki Iizuka, Masahito Yamamoto, Masashi Furukawa
    Proceedings of the Twentieth International Symposium on Artificial Life and Robotics (AROB 20th 2015) 291 - 296 2015年10月 [査読有り][通常論文]
  • Kazuma Aoyama, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    SCIENTIFIC REPORTS 5 10168  2015年05月 [査読有り][通常論文]
     
    Galvanic vestibular stimulation (GVS) can be applied to induce the feeling of directional virtual head motion by stimulating the vestibular organs electrically. Conventional studies used a two-pole GVS, in which electrodes are placed behind each ear, or a three-pole GVS, in which an additional electrode is placed on the forehead. These stimulation methods can be used to induce virtual head roll and pitch motions when a subject is looking upright. Here, we proved our hypothesis that there are current paths between the forehead and mastoids in the head and show that our invented GVS system using four electrodes succeeded in inducing directional virtual head motion around three perpendicular axes containing yaw rotation by applying different current patterns. Our novel method produced subjective virtual head yaw motions and evoked yaw rotational body sway in participants. These results support the existence of three isolated current paths located between the mastoids, and between the left and right mastoids and the forehead. Our findings show that by using these current paths, the generation of an additional virtual head yaw motion is possible.
  • Yasumasa Tamura, Hiroyuki Iizuka, Masahito Yamamoto, Masashi Furukawa
    SOFT COMPUTING 19 4 891 - 899 2015年04月 [査読有り][通常論文]
     
    This study focuses on the reactive scheduling in manufacturing systems. In practical situations, the predetermined schedule often changes under the influence of the unpredictable or variable environment in the factory. The reactive scheduling (RS) is a methodology to modify the predetermined schedule to cope with variable situations without suspending the proceeding of the processes on the schedule. Therefore, the RS requires a method to improve the schedule with fast computational time. To obtain an effective method for the RS, this paper proposes a novel RS method using local clustering organization. Also, this paper examines its effectiveness by a numerical experiment. Our results suggest that the proposed method is more effective than the method using genetic algorithm.
  • 三橋 允, 飯塚 博幸, 山本 雅人
    精密工学会学術講演会講演論文集 2015 755 - 756 公益社団法人 精密工学会 2015年 
    ゲーム木探索の中でもランダムなシミュレーションを繰り返すことで近似的な評価値を得るモンテカルロ木探索法と呼ばれる手法が存在する。しかしランダム性故に現実にはまず起こり得ないシミュレーション結果も評価に含まれてしまうことがあるため、数手に渡り正確な読みが必要となる局面は苦手とされている。そこでシミュレーション結果(報酬)をミニマックス的に伝搬することで、末端の情報を活用し評価精度の向上を目指す。
  • 遠峰 孝太, 飯塚 博幸, 山本 雅人
    精密工学会学術講演会講演論文集 2015 923 - 924 公益社団法人 精密工学会 2015年 
    群で生活する動物は蟻や魚等,数多く存在する.種により群の形態は様々であるが,同じ種の群の違いは一見するとわかりにくい.本研究では,群の全体的な振る舞いの違いが機械学習によって分類可能かどうかを検証する.分類器には,画像認識において高い成果を得ている畳み込みニューラルネットワークを動画が扱えるように拡張したものを用いて動画データから群のパターンの違いを識別することが有用であるかを検証する.
  • 加藤 修, 飯塚 博幸, 山本 雅人
    精密工学会学術講演会講演論文集 2015 915 - 916 公益社団法人 精密工学会 2015年 
    カーリングは氷上のチェスと呼ばれる戦略性の高いゲームである.また,確率的要素を持ち,取り得る手と局面が無限に存在するゲームの研究材料として注目されている.近年,カーリングシミュレータが開発され,カーリングの戦略を解析することが可能となった.本研究では,実際の試合における時間的制約を考慮したゲーム木探索に基づくカーリングAIを開発し,その有効性をシミュレーションにより検証する.
  • Yasumasa Tamura, Hiroyuki Iizuka, Masahito Yamamoto
    PROCEEDINGS OF THE 18TH ASIA PACIFIC SYMPOSIUM ON INTELLIGENT AND EVOLUTIONARY SYSTEMS, VOL 2 465 - 477 2015年 [査読有り][通常論文]
     
    Job-shop scheduling problem(JSP) is one of the hardest combinatorial optimization problems. Local clustering organization (LCO) is proposed by Furukawa et al. to solve such a combinatorial optimization problems as the metaheuristic algorithm. Its effectiveness for the JSP is verified by the comparison with genetic algorithm. However, since LCO is based on the greedy search, the solution is often trapped in local minima. To improve the problem, this study proposes a novel neighborhood search method using priority rules. This paper also shows the extended LCO integrated with the search method.
  • Masahito Yamamoto, Shu Kato, Hiroyuki Iizuka
    2015 IEEE CONFERENCE ON COMPUTATIONAL INTELLIGENCE AND GAMES (CIG) 474 - 480 2015年 [査読有り][通常論文]
     
    Strategy for digital curling based on the game tree search is demonstrated. Digital curling has been developed for simulating curling which is one of the winter sports. We present a curling playing program run on the digital curling simulator. It enable us to decide the direction and the speed of the stone to be delivered for maximizing the evaluation values of resultant game states. The novel evaluation function is constructed and the effective method for reducing the simulation trials is proposed for finding the best shot on the current game state. Our experimental results reveal that our proposed program outperforms the previous rule-based program and competes with the MonteCarlo based program.
  • Kei Takada, Masaya Honjo, Hiroyuki Iizuka, Masahito Yamamoto
    ADVANCES IN COMPUTER GAMES, ACG 2015 9525 235 - 246 2015年 [査読有り][通常論文]
     
    The game of Hex was invented in the 1940s, and many studies have proposed ideas that led to the development of a computer Hex. One of the main approaches developing computer Hex is using an evaluation function of the electric circuit model. However, such a function evaluates the board states only from one perspective. Consequently, it is recently defeated by the Monte Carlo Tree Search approaches. In this paper, we therefore propose a novel evaluation function that uses network characteristics to capture features of the board states from two perspectives. Our proposed evaluation function separately evaluates the board network and the shortest path network using betweenness centrality, and combines the results of these evaluations. Furthermore, our proposed method involves changing the ratio between global and local evaluations through a support vector machine (SVM). So, it yields an improved strategy for Hex. Our method is called Ezo. It was tested against the world-champion Hex program MoHex. The results showed that our method was superior to the 2011 version of MoHex on an 11 x 11 board.
  • 青山一真, 飯塚博幸, 櫻井悟, 宮本靖久, 古川正紘, 前田太郎, 安藤英由樹
    日本バーチャルリアリティ学会論文誌 20 4 291 - 298 2015年 [査読有り][通常論文]
  • Soichiro Yokoyama, Hiroyuki Iizuka, Masahito Yamamoto
    Journal of Advanced Computational Intelligence and Intelligent Informatics 19 6 892 - 899 2015年 [査読有り][通常論文]
     
    The heuristic method we propose solves the flexible job-shop scheduling problem (FJSP) using a solution construction procedure with priority rules. FJSP is more complex than classical scheduling problems in that operations are processed on one of multiple candidate machines, one of which must be selected to get a feasible solution. The solution construction procedure with priority rules is implemented on top of the efficient existing method for solving the FJSP which consists of a genetic algorithm and a local search method. The performance of the proposed method is analyzed using various benchmark problems and it is confirmed that our proposed method outperforms the existing method on problems with particular conditions. The conditions are further investigated by applying the proposed method on newly created benchmark.
  • Iizuka, H, Nakai, H, Yamamoto, M
    Proceedings of the European Conference on Artificial Life 264 - 270 2015年 [査読有り][通常論文]
  • Zhuoran Xu, Hiroyuki Iizuka, Masahito Yamamoto
    Artificial Life and Robotics 19 4 317 - 327 2014年12月10日 [査読有り][通常論文]
     
    In our previous work, we proposed a niching genetic algorithm: attraction basin sphere estimating genetic algorithm (ABSEGA). It can detect and recognize attraction basin spheres on a fitness landscape. Attraction basin spheres are used to improve the performance of niching. However, ABSEGA does not work well on Neuroevolution problems, because those problems are usually high dimensional and have numerous local optima. In this paper, we calculate the height of valleys to improve the niching results. The height is used to determine if an optimum is important, so we can track important optima even among hundreds of unimportant optima. We examine our method in a robotic arm problem, in which we evolve an artificial neural network to control the arm to catch balls. The results show that our method has a high ability to escape local optima and find relatively better solutions.
  • Ming Chang, Hiroyuki Iizuka, Yasushi Naruse, Hideyuki Ando, Taro Maeda
    SCIENTIFIC REPORTS 4 6729  2014年10月 [査読有り][通常論文]
     
    Neurofeedback is a strong direct training method for brain function, wherein brain activity patterns are measured and displayed as feedback, and trainees try to stabilize the feedback signal onto certain desirable states to regulate their own mental states. Here, we introduce a novel neurofeedback method, using the mismatch negativity (MMN) responses elicited by similar sounds that cannot be consciously discriminated. Through neurofeedback training, without participants' attention to the auditory stimuli or awareness of what was to be learned, we found that the participants could unconsciously achieve a significant improvement in the auditory discrimination of the applied stimuli. Our method has great potential to provide effortless auditory perceptual training. Based on this method, participants do not need to make an effort to discriminate auditory stimuli, and can choose tasks of interest without boredom due to training. In particular, it could be used to train people to recognize speech sounds that do not exist in their native language and thereby facilitate foreign language learning.
  • Tom Froese, Hiroyuki Iizuka, Takashi Ikegami
    FRONTIERS IN PSYCHOLOGY 5 1061  2014年09月 [査読有り][通常論文]
     
    According to the enactive approach to cognitive science, perception is essentially a skillful engagement with the world. Learning how to engage via a human-computer interface (HCI) can therefore be taken as an instance of developing a new mode of experiencing. Similarly, social perception is theorized to be primarily constituted by skillful engagement between people, which implies that it is possible to investigate the origins and development of social awareness using multi-user HCIs. We analyzed the trial-by-trial objective and subjective changes in sociality that took place during a perceptual crossing experiment in which embodied interaction between pairs of adults was mediated over a minimalist haptic HCI. Since that study required participants to implicitly relearn how to mutually engage so as to perceive each other's presence, we hypothesized that there would be indications that the initial developmental stages of social awareness were recapitulated. Preliminary results reveal that, despite the lack of explicit feedback about task performance, there was a trend for the clarity of social awareness to increase over time. We discuss the methodological challenges involved in evaluating whether this trend was characterized by distinct developmental stages of objective behavior and subjective experience.
  • Takumi Yokosaka, Hiroyuki Iizuka, Tomoko Yonemura, Daisuke Kondo, Hideyuki Ando, Taro Maeda
    SCIENTIFIC REPORTS 4 6201  2014年08月 [査読有り][通常論文]
     
    We report a novel illusion whereby people perceive both congruent and incongruent hand motions as a united, single, and continuous motion of one's own hand (i.e. a sense of agency). This arises when individuals watch congruent and incongruent hand motions alternately from a first person perspective. Despite an individual knowing that s/he is not performing the motion, this illusion still can arise. Although a sense of agency might require congruency between predicted and actual movements, united motion is incongruent with predicted movement because the motion contains oscillating movement which results from switching hand movement images. This illusion offers new insights into the integration mechanism of predicted and observed movements on agency judgment. We investigated this illusion from a subjective experience point of view and from a motion response point of view.
  • Soichiro Yokoyama, Hiroyuki Iizuka, Masahito Yamamoto
    2014 JOINT 7TH INTERNATIONAL CONFERENCE ON SOFT COMPUTING AND INTELLIGENT SYSTEMS (SCIS) AND 15TH INTERNATIONAL SYMPOSIUM ON ADVANCED INTELLIGENT SYSTEMS (ISIS) 749 - 754 2014年 [査読有り][通常論文]
     
    A heuristic method for solving flexible job-shop scheduling problem (FJSP) is proposed. FJSP is more complex than classical scheduling problems in that operations are processed on one of multiple candidate machines and one of them must be selected to obtain a feasible solution. The proposed method utilizes a solution improvement procedure with priority rules. This procedure is implemented on top of the efficient existing method for the FJSP which consists of genetic algorithm and local search method. The performance of the proposed method is analyzed on various benchmark problems. As a result, it is confirmed that the proposed method outperforms the existing method on problems with particular conditions while its average performance does not match the state-of-the-art algorithm.
  • Masaya Honjo, Hiroyuki Iizuka, Masahito Yamamoto
    2014 JOINT 7TH INTERNATIONAL CONFERENCE ON SOFT COMPUTING AND INTELLIGENT SYSTEMS (SCIS) AND 15TH INTERNATIONAL SYMPOSIUM ON ADVANCED INTELLIGENT SYSTEMS (ISIS) 755 - 760 2014年 [査読有り][通常論文]
     
    Insertion-based Particle Swarm Optimization (IPSO) is an optimization method for Traveling Salesman Problem (TSP) s. IPSO can find a better solution in a short time compared with Simulated Annealing (SA) and Genetic Algorithm (GA). In order to find a near optimal solution, we propose IPSOLB by improving IPSO. IPSOLB is based on the local version of PSO. In this method, a particle is affected by its neighbor particles instead of the whole population. As a result of this modification, agents are not converged immediately on the best solution obtained so far, and the population maintains various candidate solutions. For this reason, in comparison with IPSO, IPSOLB can be expected to search a solution while avoiding local optimum solutions. Through numerical experiments, it is shown that the proposed method can find a better solution than IPSO in reasonable time.
  • Ryota Kato, Hiroyuki Iizuka, Masahito Yamamoto
    2014 IEEE/SICE INTERNATIONAL SYMPOSIUM ON SYSTEM INTEGRATION (SII) 112 - 117 2014年 [査読有り][通常論文]
     
    This paper shows the different types of swarm behaviors of simulated flapping birds obtained in our simulation. The individual birds are modeled in the physical virtual environment and flying by flapping. The evolutionary method is applied to obtain how to fly as an individual. After the evolution, the multiple birds interact with each other with Kuramoto model while flying at the same time. Our result shows the top-down regulation from the shape of formations to the individual behaviors. The regulation changes individual objectives to the globally determined objectives due to the entrainment caused by Kuramoto model.
  • Ming Chang, Hiroyuki Iizuka, Yasushi Naruse, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 28:1-28:2  2014年 [査読有り][通常論文]
     
    There are a lot of skills that it takes time for us to learn in our life. To be precise, it is not clear what and how to learn. For example, one of the biggest problems in the language learning is that learners cannot recognize novel sounds that do not exist in their native language, and it is difficult to gain a listening ability for these novel sounds [1]. Here, we developed a novel neurofeedback (NF) method, using the mismatch negativity (MMN) responses elicited by similar sounds, that can help people to unconsciously improve their auditory perceptual skills. In our method, the strength of the participants' MMN as a measure of perceptual discriminability is presented as visual feedback to provide a continuous, not binary, cue for learning. We found evidence that significant performance improvement for behavioral auditory discrimination and neurophysiological measure occurs unconsciously. Based on our findings, the method has great potential to provide effortless auditory perceptual training and develop an unconscious learning interface device.
  • Ming Chang, Hiroyuki Iizuka, Yasushi Naruse, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 109 - 112 2014年 [査読有り][通常論文]
     
    There are a lot of skills that it takes time for us to learn in our life. To be precise, it is not clear what and how to learn. For example, one of the biggest problems in the language learning is that learners cannot recognize novel sounds that do not exist in their native language, and it is difficult to gain a listening ability for these novel sounds [1]. Here, we developed a novel neurofeedback (NF) method, using the mismatch negativity (MMN) responses elicited by similar sounds, that can help people to unconsciously improve their auditory perceptual skills. In our method, the strength of the participants' MMN as a measure of perceptual discriminability is presented as visual feedback to provide a continuous, not binary, cue for learning. We found evidence that significant performance improvement for behavioral auditory discrimination and neurophysiological measure occurs unconsciously. Based on our findings, the method has great potential to provide effortless auditory perceptual training and develop an unconscious learning interface device.
  • Soichiro Yokoyama, Hiroyuki Iizuka, Masahito Yamamoto
    JOURNAL OF ADVANCED MECHANICAL DESIGN SYSTEMS AND MANUFACTURING 8 5 JAMDSM0073  2014年 [査読有り][通常論文]
     
    We propose an efficient heuristic method for job-shop scheduling problems (JSP) with the objective of total weighted tardiness minimization The proposed method uses schedule reconstructions by priority rules to guide a local search towards promising solutions. Typically priority rules determine the whole schedule and thus incorporating it with local search procedure is difficult. In our proposed method, a priority rule decides a schedule within an arbitrary selected time window and the rest of the schedule is determined by the schedule obtained by a conventional local search method. The priority rule is given by a linear combination of simple priority rules. To improve a schedule efficiently, an appropriate set of time window and gains of the linear combined priority rule is required. Therefore, we optimize the set of time window and rule gains with genetic algorithm This rule-based reconstruction procedure and the conventional local search procedure are alternately applied to a current solution. In the experiments, the efficiency of rule-based reconstruction procedure is verified and the proposed method is compared with one of the most effective existing methods. The results show that the proposed method outperforms the existing method on large problems with sufficient computational time. The average performance is particularly improved due to the ability of rule-based reconstruction procedure to escape from local optima of the conventional local search procedure efficiently.
  • ネットワーク指標を用いたコンピュータ Hex 戦略の開発
    高田圭, 本庄将也, 飯塚博幸, 山本雅人
    情報処理学会論文誌 55 11 2421 - 2430 2014年 [査読有り][通常論文]
  • 青山一真, 安藤英由樹, 飯塚博幸, 前田太郎
    日本バーチャルリアリティ学会論文誌 19 3 315 - 318 特定非営利活動法人 日本バーチャルリアリティ学会 2014年 [査読有り][通常論文]
     
    In this work, we researched the human perceptual attribution for acceleration sensation evoked by Galvanic Vestibular Stimulation (GVS) and validated the enhancement effect of countercurrent using the dead current strength. In the results of experiments, we searched that the current threshold is 0.3 - 0.4mA by measuring the rate of correct directional perception. To validate the enhancement effect of countercurrent, we compared the rate of correct perception in long and short countercurrent and constant current. In the result, the rate of correct directional perception in long countercurrent is significantly higher than that of constant current in the dead current value.
  • Tom Froese, Hiroyuki Iizuka, Takashi Ikegami
    SCIENTIFIC REPORTS 4 3672  2014年01月 [査読有り][通常論文]
     
    Scientists have traditionally limited the mechanisms of social cognition to one brain, but recent approaches claim that interaction also realizes cognitive work. Experiments under constrained virtual settings revealed that interaction dynamics implicitly guide social cognition. Here we show that embodied social interaction can be constitutive of agency detection and of experiencing another's presence. Pairs of participants moved their "avatars"' along an invisible virtual line and could make haptic contact with three identical objects, two of which embodied the other's motions, but only one, the other's avatar, also embodied the other's contact sensor and thereby enabled responsive interaction. Co-regulated interactions were significantly correlated with identifications of the other's avatar and reports of the clearest awareness of the other's presence. These results challenge folk psychological notions about the boundaries of mind, but make sense from evolutionary and developmental perspectives: an extendible mind can offload cognitive work into its environment.
  • 複雑系アプローチによる他者の認識とコミュニケーションの創発
    飯塚博幸
    計測と制御 53 9 821 - 827 2014年 [査読無し][招待有り]
  • 近藤 大祐, 飯塚 博幸, 安藤 英由樹, 小濱 和貴, 坂井 義治, 前田 太郎
    日本バーチャルリアリティ学会論文誌 18 4 487 - 496 日本バーチャルリアリティ学会 2013年12月31日 [査読無し][通常論文]
     
    We propose a novel training method for laparoscopy using view-sharing technology where the expert's skillful movements are superimposed on the learner's view. When learning the skill, the learner just traces the expert's movement as if they feel the skillful movements as their own. Our proposed method was evaluated in terms of time required for the task and the number of errors. The training effect was also tested with the doctors in training and the acquired skills were evaluated by the expert surgeon.
  • Yuki Sato, Hiroyuki Iizuka, Takashi Ikegami
    CONSTRUCTIVIST FOUNDATIONS 9 1 73 - + 2013年11月 [査読有り][通常論文]
     
    Context . Our body schema is not restricted to biological body boundaries (such as the skin), as can be seen in the use of a cane by a person who is visually impaired or the "rubber hands" experiment. The tool becomes a part of the body schema when the focus of our attention is shifted from the tool to the task to be performed. Problem . A body schema is formed through interactions among brain, body, tool, and environment. Nevertheless, the dynamic mechanisms underlying changes in the body schema are still not fully understood. Method . To study the changing conditions of the body schema (e.g., a shift of attention), a simulation model of object discrimination was extended to differentiate between two kinds of sensitivities - sensitivity to an object being directly manipulated and sensitivity to another object being manipulated by the first. The proposed model consisted of windmills with different numbers of vanes. A model agent was required to determine the number of vanes on a windmill by touching the vanes blindly with an arm controlled by a neural network. Placing a second windmill beside the first and gearing the two windmills to move associatively resulted in the agent using the first windmill as a tool with which to discern the number of vanes on the second windmill. In other words, an agent's body schema can shift from its arm tip to the boundary between the first and second windmills. We then introduced an experiment with a real windmill model to test the hypothesis demonstrated by the theoretical model. Results . We demonstrated that even simple computational agents can have two different sensitivities to the windmills. One agent becomes sensitive to the first windmill and insensitive to the second one. Another agent becomes insensitive to the first windmill and sensitive to the second one by using the first one as a tool. Therefore, we concluded that the boundary of the body schema was extended to the first windmill in the case of the latter agent because paying attention to the task to be performed instead of the tool itself is essential for the tool to be considered as part of the body schema. Analysis of the experiments using a computational model and human experimentation revealed that a shift from an irregular to a regular movement of a windmill is an indication of extension of the body schema. Constructivist content . Our insights are beneficial for enactive cognitive science. This is because an extended body schema questions the Cartesian separation between subject and object, and the self and the environment.
  • Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    ADAPTIVE BEHAVIOR 21 4 263 - 273 2013年08月 [査読有り][通常論文]
     
    This study presents an extended dynamic neural network model of homeostatic adaptation as the first step toward constructing a model of mental imagery. In the homeostatic adaptation model, higher-level dynamics internally self-organized from sensorimotor dynamics are associated with desired behaviors. These dynamics are regenerated when drastic changes occur, which might break the internal dynamics. Due to the weak link between desired behavior and internal homeostasis in the original homeostatic adaptation model, adaptivity is limited. In this paper, we improve on the homeostatic adaptation model to create a stronger link between desired behavior and internal homeostasis by introducing a metabolic causation in a plasticity mechanism and show that it becomes more adaptive. Our results show that our model has three different time scales in the adaptive behaviors, which are discussed with our cognition and mental imagery.
  • 砂川 未佳, 丹羽 真隆, 飯塚 博幸, 安藤 英由樹, 前田 太郎
    日本バーチャルリアリティ学会論文誌 18 1 45 - 54 日本バーチャルリアリティ学会 2013年03月31日 [査読無し][通常論文]
     
    This study aimed at investigating if CNV-Iike variations are elicited in Electroencephalograms (EEG) when people remember human continuous motion. It is known that there are some important postures in continuous human motion, and the segments divided by the significant postures attract human attention. Since the situation is similar to Event-related brain potential (ERP) experiments, we expected that brain wave variations like Contingent Negative Variation (CNV) would also be elicited during motion remembering. In this study, we found that negative variations tend to precede segments divided by manually selected significant postures.
  • Kazuma Aoyama, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    23rd International Conference on Artificial Reality and Telexistence, ICAT 2013, Tokyo, Japan, December 11-13, 2013 116 - 121 IEEE Computer Society 2013年 [査読有り][通常論文]
  • Daisuke Kondo, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    2013 IEEE SYMPOSIUM ON 3D USER INTERFACES (3DUI) 145 - 146 2013年 [査読有り][通常論文]
     
    We propose a novel method to navigate one's head rotation and position when two persons perform cooperative work with wearing head-mounted display, which is called view-sharing system. It is necessary to synchronize and match their head motions between two persons. In this system, the head motion is transmitted and guided by the target-maker shown in the user's view. In a conventional method, the partner's head position was shown in the view as target marker. In this paper, the marker was modified in terms of two typical human motion patterns; changing point of view by moving one's head around the target object, and changing one's gaze point by turning the head. Our proposed method was evaluated with the head tracking task and showed the effectiveness.
  • Hiroyuki Iizuka, Mika Sunagawa, Masataka Niwa, Hideyuki Ando, Taro Maeda
    2013 IEEE VIRTUAL REALITY CONFERENCE (VR) 2013年 [査読有り][通常論文]
     
    Our aim of this study is to investigate the possibility of estimating interruptibility from Electroencephalograms (EEG). We focus on contingent negative variation which is elicited in response to concentration. Our results support that our method could correctly detects the user's internal states such as just watching or actively remembering even the external situations were completely same and also CNV is elicited when controlling continuous motion.
  • Hiroyuki Iizuka, Davide Marocco, Hideyuki Ando, Taro Maeda
    Psychological Research 77 1 53 - 63 2013年01月 [査読有り][通常論文]
     
    In this paper, we present an experiment that integrates a semiotic investigation with a dynamical perspective on embodied social interactions. The primary objective is to study the emergence of a communication system between two interacting individuals, where no dedicated communication modalities are predefined and the only possible interaction is very simple, non-directional, and embodied. Throughout the experiment, we observe the following three phenomena: (1) the spontaneous emergence of turn-taking behaviour that allows communication in non-directional environments (2) the development of an association between behaviours and perceptive categories (3) the acquisition of novel meaning by exploiting the notion of complementary set theory. © 2012 Springer-Verlag.
  • 橋本 悠希, 杉崎 有, 米村 朋子, 飯塚 博幸, 安藤 英由樹, 前田 太郎
    日本バーチャルリアリティ学会論文誌 17 1 23 - 32 日本バーチャルリアリティ学会 2012年03月31日 [査読無し][通常論文]
     
    Galvanic vestibular stimulation is known to cause ocular movement. However, the effects of GVS on ocular movements have only been investigated when gazing at a certain point, despite the fact that humans use two different strategies to follow moving targets: saccade and smooth pursuit. The effects of GVS may be different for these two strategies. This paper investigates the effects of GVS during saccade and smooth pursuit. The results show that GVS exerts different effects on each of these eye movement strategies.
  • 丹羽 真隆, 飯塚 博幸, 安藤 英由樹, 前田 太郎
    日本バーチャルリアリティ学会論文誌 17 1 3 - 10 日本バーチャルリアリティ学会 2012年03月31日 [査読無し][通常論文]
     
    We propose novel architecture to realize a low-degree of freedom (DOF) Tsumori controller to manipulate a high-DOF robot. We believe that humans can immerse themselves to manipulate a robot if the robot reflects user intention. In our Tsumori controller, the user intuitively operates a joystick, and the robot moves semi -autonomously and reflects the user's intention that is extracted from intuitive input sequences. In this paper, we explain how to extract Tsumori, which is an archetype of human behavioral intention, and how to realize a Tsumori controller.
  • Yoshito Kumakura, Yuki Hashimoto, Daisuke Kondo, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    2012 IEEE/SICE INTERNATIONAL SYMPOSIUM ON SYSTEM INTEGRATION (SII) 410 - 415 2012年 [査読有り][通常論文]
     
    When human exercises multi-degrees of freedom (DOF) motion, they just pay attention to only part of the body motion and can control the whole body motion. By consciously adjusting low-DOF movements, it is possible to induce multi-DOF motion, which is referred as a "knack" in the sport. Acquiring knacks can drastically improve multi-DOF motor skills by low-DOF information presentation, and it is also used in motor learning. We consider that "knack" is "what induces multi-DOF motion by low-DOF information presentation" and "what improves motor skills". In this study, we investigated whether multi-DOF motion can be induced by single-DOF information presentation, that is sound frequency, and whether motor skills are improved through our human experiments.
  • Hiroyuki Iizuka, Davide Marocco, Hideyuki Ando, Taro Maeda
    IEEE VIRTUAL REALITY CONFERENCE 2012 PROCEEDINGS 1 - 4 2012年 [査読有り][通常論文]
     
    Our aim of this paper is to investigate the human communication in terms of two following questions. One is how human can know the fact that an interacting partner is human. Another is how non-communicative behaviour can become communicative. To answer these questions, we performed two experiments exploiting the idea of perceptual crossing experiments. As a result, we will show that the turn-taking structure supports humanlikeness and human communication in the primitive non-verbal interaction. Our results will be discussed with ambient interface technology.
  • Hidenori Hamaguchi, Masataka Niwa, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    IEEE VIRTUAL REALITY CONFERENCE 2012 PROCEEDINGS 1 - 2 2012年 [査読有り][通常論文]
     
    We developed a new pseudo-attraction force display using four vibrating motors. The advantage of this device is that it uses phase difference control to alter the time width of pulses to generate asymmetric oscillation. It does not require altering the frequency of the vibrating motors. In this study, we investigated the influence of the pulse width generated in the device using vibrating motors on the perceptive intensity of pseudo-attraction force. From the result, the oscillation having shorter pulse widths within the range of 10-20 ms induced the "being-pulled" sensation more strongly.
  • Yuki Hashimoto, Daisuke Kondo, Tomoko Yonemura, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    ACM SIGGRAPH 2012 Posters, SIGGRAPH'12 115  2012年 [査読有り][通常論文]
  • Hideyuki Ando, Masataka Niwa, Hiroyuki Iizuka, Taro Maeda
    Proceedings - IEEE International Workshop on Robot and Human Interactive Communication 304 - 307 2012年 [査読有り][通常論文]
     
    When training for dance or sports, amateurs observe and mimic the physical exertion of experts to acquire skill by observing others. However, humans don't memorize consecutive operations, but instead make discrete divisions and memorize body movements. The posture that becomes a key from consecutive operations is removed and memorized for discretion in a process we call segmentation. The movement is reproduced by smooth interpolations between the clause and the clause memorized as segmentation. In the first experiment, we clarified the existence of this segmentation by showing video of smooth movements to participants who selected the segment/frame that became a key. We confirmed commonness in this segment selection. In experiment 2, to reflect the physical exertion that made the segmentation, the participants input force to force sensors by a control stick. We showed video of smooth movements to them and simultaneously input the force pattern that corresponds to the movement. The force pattern, which uniformly converges by such repetition, is different depending on each participant. We call this method the extraction of segmented behavioral intention. On the other hand, we can arbitrarily change the speed of the physical exertion by expanding and contracting the segments of the movement video that obtained the force pattern by extraction of segmented behavioral intention. Since the expanded and contracted pattern range at a certain speed even when the movement speed changed, we preserved the shape and the selected segments. However, the segment shape changes when the movement speed is constantly fast. © 2012 IEEE.
  • Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda, Davide Marocco
    Proceedings - IEEE International Workshop on Robot and Human Interactive Communication 298 - 303 2012年 [査読有り][通常論文]
     
    This study presents an experiment that integrates a semiotic investigation with a dynamical perspective on embodied social interactions. The primary objective is to study co-creativity of a communication system on the human-human nonverbal interaction. Throughout the experiment, the results of established communication system on uni- and bi-directional interactions are compared. Finally, the role of co-creativity in the evolution of communication systems will be discussed. © 2012 IEEE.
  • 飯塚 博幸, 安藤 英由樹, 前田 太郎
    電子情報通信学会論文誌. A, 基礎・境界 95 1 165 - 174 一般社団法人電子情報通信学会 2012年01月01日 [査読無し][通常論文]
     
    人と人工システムの調和的な相互作用の成立を目指し,本論文では人同士が行う相互行為からなるコミュニケーションに焦点を当てる.言語コミュニケーションを含む日常の人同士の相互作用は多くのモダリティを含み,解析が困難なものとなり,かつ,我々が日常使っているために多くの先入観を含み対象として適さない.そのため,相互作用を簡単な身体の運動と知覚に十分に制限し,2者間においてあらかじめ意味が与えられていない身体的相互作用からコミュニケーションが成立していく過程を調べることによって,人同士のコミュニケーションの背後にある身体的相互作用のダイナミクスを明らかにする.実験では,相手が人かどうかを判断することだけを目的にしているにもかかわらず,人の言語コミュニケーションに見られる時間的役割の転換であるターンテイキングが,身体的相互作用に創発することを示す.
  • 橋本 悠希, 熊倉 祥人, 米村 朋子, 飯塚 博幸, 安藤 英由樹, 前田 太郎
    日本バーチャルリアリティ学会論文誌 16 4 633 - 641 日本バーチャルリアリティ学会 2011年12月31日 [査読無し][通常論文]
     
    We sometimes experience a positive feeling during playing sports/exercising without knowing the consequences of the motions. The feeling is called "Shikkuri", which is an onomatopoeia in Japanese which is considered an important feeling when learning motion skills. Some coaches already use this onomatopoeia in motion skill training. However it is not clear at all how and when the feeling arises. Our aim is to clarify the mechanism of the feeling and to exploit the feeling for teaching motion skills. As the first step toward our goal, we identify the factors related to perception of the Shikkuri feeling in the case of pitching.
  • 米村 朋子, 橋本 悠希, 近藤 大祐, 丹羽 真隆, 飯塚 博幸, 安藤 英由樹, 前田 太郎
    日本バーチャルリアリティ学会論文誌 16 4 623 - 632 日本バーチャルリアリティ学会 2011年12月31日 [査読無し][通常論文]
     
    This paper investigates the learning effects of the Cardio Pulmonary Resuscitation (CPR) using the view-sharing system that can blend the views from the first person perspective of the instructor and the learner. The usual way of learning the CPR procedures is watching the video of instructor's action recorded from the third person perspective and trying CPR. However, there are difficulties for learners in the third person perspective video to recognize where the trainer pays attention and what he/she checks. Therefore, we clarify how the first person perspective helps for learning of CPR. As a result, we found that learners wearing our view-sharing system make better performances than the conventional method and can easily recognize their mistakes of procedures. We discuss about effects of skill training using the first person perspective by the view-sharing system.
  • 丹羽 真隆, 濱口 英典, 飯塚 博幸, 前田 太郎, 安藤 英由樹
    日本バーチャルリアリティ学会論文誌 16 3 409 - 414 日本バーチャルリアリティ学会 2011年09月30日 [査読無し][通常論文]
     
    The paper studies two methods of haptic. One is an attraction force display that can induce attraction force by a cyclic movement of a weight with asymmetric acceleration. Another is a texture display that induces a sensation of texture by giving off vibration while a subject traces a finger over a flat surface. In this paper, we propose a novel design that can induce both attraction force and texture feeling. The prototype consists of four vibration motors, which are controlled to generate asymmetric acceleration and vibration. The device is evaluated by experiments.
  • 橋本 悠希, 湯村 武士, 米村 朋子, 飯塚 博幸, 前田 太郎, 安藤 英由樹
    日本バーチャルリアリティ学会論文誌 16 3 399 - 408 日本バーチャルリアリティ学会 2011年09月30日 [査読無し][通常論文]
     
    We propose a method to transmit tactile information from one person to another using nail-mounted tactile display. Our proposed method conveys tactile information that is obtained by detecting the vibration on the nail of a sender when tracing a texture, and generates a vibration on the nail of a receiver to re-create the tactile perception while the subject tracing on a flat surface. In this paper, we reported about measuring method of tactile information, control method of actuator and developing a new actuator in order to share tactile perceptions with two people.
  • Shin Okamoto, Hiroki Kawasaki, Hiroyuki Iizuka, Takumi Yokosaka, Tomoko Yonemura, Yuki Hashimoto, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 32  2011年 [査読有り][通常論文]
     
    This paper reports a study of augmenting human motions by manipulating visual images displayed to users. The target motion is not only the motion that can be seen in the subject's views (i.e. hands or foots motion) but also the full-body motion that cannot be captured from their own perspective. As a result, it is shown that the motions are modulated without any physical contacts only by manipulated images. Copyright c 2011 ACM.
  • Aru Sugisaki, Yuki Hashimoto, Tomoko Yonemura, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 28  2011年 [査読有り][通常論文]
     
    It is known that galvanic vestibular stimulation can cause ocular movement. Our final goal is to use GVS to support ocular movements. However, the effects of GVS to ocular movements are basically investigated while gazing at a certain point despite the fact that we have two different strategies to follow a moving target such as saccade and smooth pursuit. The effect might be different because those two use different mechanism. Therefore, this paper investigates the GVS effects during saccade. As a result, we show that the effect of GVS depends on the timing when GVS is given after the target marker moves. Copyright © 2011 ACM.
  • Daisuke Kondo, Keitaro Kurosaki, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 26  2011年 [査読有り][通常論文]
     
    We are developing 'view sharing' system for supporting a remote corporative work. The view sharing is constructed from the video-see-through head mounted displays (VST-HMD) and motion trackers. This system allows two users in remote places to share their first-person views each other. The users can share what the other user is seeing, and furthermore the users can correspond their spatial perception, motion and head movement. By sharing those sensations, the non-verbal skills can be transmitted from skilled person to the non-skilled person. Using this system expert in remote place can instruct the non-skilled person to improve task performance. Copyright © 2011 ACM.
  • Hiroyuki Iizuka, Daisuke Kondo, Hiroki Kawasaki, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 23  2011年 [査読有り][通常論文]
     
    We describe how visually coupled people start their synchronized behavior with two visual coupling conditions: view-swapping and view-blending. In the view-swapping condition, two people's views are changed from the first-person perspective so that both see their partner's views. The view-blending condition allows people to see the blended view of both views. We report the results of different coordinated strategies to start synchronization that is observed in different conditions. In terms of the time required to start synchronization, view-swapping outperforms view-blending. Copyright © 2011 ACM.
  • Keitaro Kurosaki, Hiroki Kawasaki, Daisuke Kondo, Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    ACM International Conference Proceeding Series 20  2011年 [査読有り][通常論文]
     
    In this paper, we describe the skill transmission through our viewsharing system that can mix or exchange the first person perspectives from the exact the partner's viewpoints. Since a nonskilled person can see the first person perspective of a skilled person, the motion of the non-skilled person is intuitively modified and supported. The task for the skill transmission is to play theremin that requires precise hand motions. As a result we show that the skill transmission effectively happens with our view-sharing system compared with the conventional method, side-by-side teaching. The way of effective augmenting human ability will be discussed. Copyright © 2011 ACM.
  • Taro Maeda, Hideyuki Ando, Hiroyuki Iizuka, Tomoko Yonemura, Daisuke Kondo, Masataka Niwa
    ACM International Conference Proceeding Series 18  2011年 [査読有り][通常論文]
     
    The Parasitic Humanoid (PH) is a wearable robotic human interface for sampling, modeling, and assisting nonverbal human behavior. This anthropomorphic robot senses the behavior of the wearer and has the internal models to learn the process of human sensory motor integration, thereafter it begins to predict the next behavior of the wearer using the learned models. When the reliability of the prediction is sufficient, the PH outputs the difference from the actual behavior as a request for motion to the wearer by motion induction using sensory illusion. Through the symbiotic interaction, the internal model and the process of human sensory motor integration approximate each other asymptotically. This process is available to transmit modalities such as senses of sight, hearing, touch, force and balance with human embodiment. This synergistic multimodal communication between distant people wearing PH can realize experience-sharing, skill transmission, and human behavior supports. Copyright 2011 ACM.
  • 飯塚 博幸, 丹羽 真隆, 近藤 大祐, 安藤 英由樹, 前田 太郎
    情報処理 51 7 767 - 774 一般社団法人情報処理学会 2010年07月 [査読無し][通常論文]
     
    ユーザの身体性を考慮し,自然かつ直観的に感覚や意図を伝送するためのインタフェースを紹介する.本稿では,従来のインタフェースでは表現することの困難な空間知覚,運動感覚やスキル,意図を取り上げる.感覚を伝送するためには運動レベルでの作用が必要であり,いかにして運動に働きかけることのできるインタフェースを実現するかを解説する.また,ヒトの意図は直観的な振舞いにおける運動に現れるため,それを検出する手法について述べる.
  • Hiroki Kawasaki, Hiroyuki Iizuka, Shin Okamoto, Hideyuki Ando, Taro Maeda
    2010 IEEE RO-MAN 125 - 131 2010年 [査読有り][通常論文]
     
    When two distant persons establish cooperative interaction, how to share their motions and sensations and how to adjust or revise their motions are important for cooperation. In this paper, we propose an approach for sharing first-person perspectives to establish collaboration and to transmit a skill from one to another. We developed a view-sharing system to realize such interaction. To investigate the fundamental behavioral property of humans with our system, we examined a simple collaborated behavior and found that a situation where subjects only see their partner's view improves velocity following. By exploiting this property, we transmitted the basic skill required for the Diabolo trick to non-skilled persons. Through our system, the performances of non-skilled persons with the assistance of a skilled person surpassed individual performances
  • 前田 太朗, 安藤 英由樹, 飯塚 博幸
    日本バーチャルリアリティ学会誌 = Journal of the Virtual Reality Society of Japan 14 2 99 - 103 2009年06月30日 [査読無し][通常論文]
  • 山下 晃弘, 川村 秀憲, 飯塚 博幸
    情報処理学会論文誌 論文誌トランザクション 2008 2 47 - 57 情報処理学会 2009年04月 [査読無し][通常論文]
  • 山下 晃弘, 川村 秀憲, 飯塚 博幸, 大内 東
    情報処理学会論文誌. 数理モデル化と応用 2 2 47 - 57 一般社団法人情報処理学会 2009年03月27日 [査読無し][通常論文]
     
    Web 上の情報量拡大にともなって,オンラインストアなどでは推薦システムを導入する事例が増加している.同時に,これまで様々な推薦アルゴリズムが提案,分析,評価されてきた.システムによる推薦結果は,次にユーザが閲覧するアイテムとなり,結果としてどのアイテムが次に評価されるかに影響を及ぼす.したがって,推薦システムとユーザは相互作用を及ぼしあう関係にあるといえる.しかし,先行研究による推薦アルゴリズムの評価・分析では,一時的な推薦精度のみに着目し,このような相互作用は考慮されてこなかった.そこで本研究では,ユーザをエージェントとしてモデル化することで,推薦システムとユーザの相互作用を考慮した,新たな評価・分析モデルを提案する.さらに,本論文では,基本的な推薦アルゴリズムである 「ユーザ間協調フィルタリング」 と,実システムとしてすでに実用化されている 「アイテム間協調フィルタリング」 を例として取り上げ,その特性について提案モデルを用いて分析した.その結果,既存の評価・分析モデルの枠を超え,より有効な推薦への指針を検証可能であることが示された.
  • Hidenori Kawamura, Hiroyuki Iizuka, Toshihiko Takaya, Azuma Ohuchi
    Artificial Life and Robotics 13 2 504 - 507 2009年 [査読有り][通常論文]
     
    We report on the cooperative control of multiple neural networks for an indoor blimp robot. In our research group, the indoor blimp robot has been studied to achieve various flying robot applications. The objective of this article is to propose a robust controller that can adapt to mechanical accidents such as the breakdown of propellers. In our proposed method, each propeller thrust is independently calculated by a small neural network. We confirm the advantage of the proposed method against the control by a single large neural network. © International Symposium on Artificial Life and Robotics (ISAROB). 2009.
  • Hiroyuki Iizuka, Hideyuki Ando, Taro Maeda
    HUMAN-COMPUTER INTERACTION, PT III 5612 284 - 293 2009年 [査読有り][通常論文]
     
    This paper proposes the concept of Parasitic Humanoid (PH) that can realize a wearable robot to establish intuitive interactions with wearers rather than conventional counter-intuitive ways like key-typing. It requires a different paradigm or interface technology which is called behavioral or ambient interface that can harmonize human-environment interactions to naturally lead to a more suitable state with the integration of information science and biologically inspired technology. We re-examine the use of wearable computers or devices from the viewpoint of behavioral information. Then, a possible way to realize PH is shown as integrated wearable interface devices. In order that PH establishes the harmonic interaction with wearers, a mutually anticipated interaction between a computer and human is necessary. To establish the harmonic interaction, we investigate the social interaction by experiments of human interactions where inputs and outputs of subjects are restricted in a low dimension at the behavioral level. The results of experiments are discussed with the attractor superimposition. Finally, we will discuss integrated PH system for human supports.
  • Taro Maeda, Hideyuki Ando, Hiroyuki Iizuka
    ACM International Conference Proceeding Series 81 - 88 2009年 [査読有り][通常論文]
     
    The Parasitic Humanoid (PH) is a wearable robot for modeling nonverbal human behavior. This anthropomorphic robot senses the behavior of the wearer and has the internal models to learn the process of human sensory motor integration, thereafter it begins to predict the next behavior of the wearer using the learned models. When the reliability of the prediction is sufficient, the PH outputs the errors from the actual behavior as a request for motion to the wearer. Through symbiotic interaction, the internal model and the process of human sensory motor integration approximate each other asymptotically. Copyright 2009 ACM.
  • Simulation and implement of memory-based PID control for indoor blimp robot
    Yoshifumi Yamagata, Hidenori Kawamura, Azuma Ohuchi, Toshihiko Takaya, Hiroyuki Iizuka
    Proceedings of the 13th International Symposium on Artificial Life and Robotics, AROB 13th'08 151 - 154 2008年 
    We report on simulation and implement of Memory-based PID control for indoor blimp robot. In the control of indoor blimp robot, it is known that the PID control is effective. However, it is necessary to adjust PID parameters dynamically, since the environment changes dynamically. We use Memory-based PID controller for PID parameter tuning. The method is that a set of the PID parameters and input/output data of the controlled object are stored in the data-base, and are used to calculate the PID parameters to control object. The results of experiments demonstrated the effectiveness of Memory-based PID control for indoor blimp robot. ©ISAROB 2008.
  • Cooperative control of multiple neural networks for indoor blimp robot
    Ryouta Nishioka, Hidenori Kawamura, Azuma Ohuchi, Toshihiko Takaya, Hiroyuki Iizuka
    Proceedings of the 13th International Symposium on Artificial Life and Robotics, AROB 13th'08 131 - 134 2008年 
    We report on cooperative control of multiple neural networks for indoor blimp robot. In our laboratory, the indoor blimp has been studied to achieve various applications. Our objective of this paper is to propose a robust controller that can adapt to mechanical accidents such as breakdown of propellers. In our proposal method, each propeller thrust is independently calculated by each small network. We confirm effectiveness of the proposed method compared to the method of calculating thrusts by a single large neural network on the simulator. ©ISAROB 2008.
  • Takashi Ikegami, Hiroyuki Iizuka
    INFANT BEHAVIOR & DEVELOPMENT 30 2 278 - 288 2007年05月 [査読有り][通常論文]
     
    In the present article, computer simulation of turn-taking interaction was studied and compared with psychological experiments such as the double TV experiment and synchronous imitation. By introducing the concepts of virtual agents, prediction error and adaptability, the simulation showed that turn-taking dynamics is a cooperative and co-creative process between two agents, which is in accordance with psychological findings. In particular, much attention is paid to styles of motion, as turn-taking with chaotic spatial pattern has developed to be sensitive to it. (c) 2007 Elsevier Inc. All rights reserved.
  • Takashi Ikegami, Hiroyuki Iizuka
    Imitation and Social Learning in Robots, Humans and Animals: Behavioural, Social and Communicative Dimensions 301 - 322 2007年01月01日 [査読有り][通常論文]
     
    We introduce a simulation study designed to develop a “synthetic psychology” (Braitenberg, 1984). Instead of studying real human/animal behavior, we study autonomous behavior of mobile agents that have virtual sensors to “see” the world and move around with virtual “wheels.” The autonomous behavior is synthesized by the agent's internal dynamics, which evolve using a genetic algorithm. A purpose of synthetic psychology is to create a new concept in order to understand an essential psychological phenomenon. Therefore, the simulation framework has to use models sufficiently simple to analyze the behavior, but sufficiently complex to provide a theory to understand the psychological experiments. As an example of synthetic psychology, we describe a simulation study of turn-taking using coupled dynamical recognizers. Studying turn-taking phenomena in robot experiments (e.g. Brooks et al., 1999 Dautenhahn, 1999 Steels and Kaplan, 2001 Miyake et al., in press) stresses the importance of embodiment and situatedness. In particular, MIT AI Lab's well-designed face robot, Kismet (Kismet project, 2000), can make use of human social protocols to stimulate natural emotions and social responses in humans. Even conversational turn-taking is established with a human subject. To do this, Kismet relies on subtle cues of human expression and interaction dynamics. Dautenhahn et al. used a mobile robot and a small humanoid robot to investigate the use of robotic toys in therapy and education of children with autism (Dautenhahn et al., 2002 AuRoRA project, 2005). These robots engage in playing a game with autistic children to elicit turn-taking and imitation.
  • Hiroyuki Iizuka, Ezequiel A. Di Paolo
    ADAPTIVE BEHAVIOR 15 4 359 - 376 2007年 [査読有り][通常論文]
     
    A preference is not located anywhere in the agent's cognitive architecture, but it is rather a constraining of behavior which is in turn shaped by behavior. Based on this idea, a minimal model of behavioral preference is proposed. A simulated mobile agent is modeled with a plastic neurocontroller, which holds two separate high dimensional homeostatic boxes in the space of neural dynamics. An evolutionary algorithm is used for creating a link between the boxes and the performance of two different phototactic behaviors. After evolution, the agent's performance exhibits some important aspects of behavioral preferences such as durability and transitions. This article demonstrates (1) the logical consistency of the multi-causal view by producing a case study of its viability and providing insights into its dynamical basis and (2) how durability and transitions arise through the mutual constraining of internal and external dynamics in the flow of alternating high and low susceptibility to environmental variations. Implications for modeling autonomy are discussed.
  • Akihiro Yamashita, Hidenori Kawamura, Hiroyuki Iizuka, Azuma Ohuchi
    Intelligent Data Engineering and Automated Learning - IDEAL 2007(IDEAL) 1112 1121 1112 - 1121 2007年 [査読有り][通常論文]
  • Emergence of body image and dichotomy of sensory and motor activity
    Hiroyuki Iizuka, Takashi Ikegami
    AISB'05 Convention: Proceedings of the Symposium on Next Generation Approaches to Machine Consciousness: Imagination, Development, Intersubjectivity and Embodiment 104 - 109 2005年 
    This paper reconsiders a boundary between an agent and its environment. The boundary between a simulated agent and its environment is convoluted in dynamic processes of sensory and motor devices. However the boundary is not a static but a dynamic interface. In order to study the dynamic property of the interface, the un-fixed sensory-motor distinction is introduced. Practically, a mobile arm passively or actively explores an object and attaining some discrimination tasks. We develop the discrimination ability by using a genetic algorithm. A conscious state towards an object is investigated with this framework.
  • Hiroyuki Iizuka, Takashi Ikegami
    Connection Science 16 4 283 - 299 2004年12月 [査読有り][通常論文]
     
    To study the interfacial complexity between an agent and its environment such as the adaptive selection aspects of sensory inputs, we propose a new coupling mechanism, called autonomous coupling, where an agent can spontaneously switch on or off its interaction with the environment. An oscillatory neural system with autonomous coupling sums the sensory inputs and initiates action selection via a sensorimotor coupling. An example task we designed to show dynamical categorization is the classification of light frequencies. An evolved agent selects specified light frequencies by approaching them and avoiding light of other frequencies. Dynamical categorization and active coupling are the key concepts for the understanding of situated and embodied cognitive functions.
  • Hiroyuki Iizuka, Takashi Ikegami
    Artificial Life 10 4 361 - 378 2004年09月 [査読有り][通常論文]
     
    Turn-taking behavior is simulated in a coupled-agents system. Each agent is modeled as a mobile robot with two wheels. A recurrent neural network is used to produce the motor outputs and to hold the internal dynamics. Agents are developed to take turns on a two-dimensional arena by causing the network structures to evolve. Turn taking is established using either regular or chaotic behavior of the agents. It is found that chaotic turn takers are more sensitive in response to inputs from the other agent. Conversely, regular turn takers are comparatively robust against noisy inputs, owing to their restricted dynamics. From many observations, including turn taking with virtual agents, we claim that there is a complementary relationship between robustness and adaptability. Furthermore, by investigating the recoupling of agents from different GA generations, we report the emergence of a new turn-taking behavior. Potential for synthesizing a new form of interaction is another characteristic of chaotic turn takers. © 2004 Massachusetts Institute of Technology.
  • Hiroyuki Iizuka, Takashi Ikegami
    Lecture Notes in Artificial Intelligence (Subseries of Lecture Notes in Computer Science) 2801 336 - 345 2003年 
    Turn-taking behaviour is simulated with a coupled agents system. Each agent is modelled as a mobile robot with two wheels. A recurrent neural network is used to produce the motor outputs and to hold the internal dynamics. Agents are developed to take turns on a two dimensional arena by causing the network structures to evolve. Turn-taking is established using either regular or chaotic behaviour of the agents. It is found that chaotic turn-takers are more sensitive to the adaptive inputs from the other agent. On the other hand, regular turntakers are comparatively insensitive to noisy inputs due to their restricted dynamics. From various observations, including turn-taking with virtual agents, we claim that the chaotic turn-taking agents become less robust when coping with virtual agents but at the same time, those agents are more adaptable to each other than the regular turn-taking agents. All these findings are discussed and compared with Trevarthen's double monitor experiments.
  • Hiroyuki Iizuka, Masatoshi Hiramoto, Hidenori Kawamura, Masahito Yamamoto, Azuma Ohuchi
    RoboCup 2000: Robot Soccer World Cup IV 449 - 452 Springer 2000年 [査読有り][通常論文]
  • Learning of words in negotiation between autonomous agents
    Hiroyuki Iizuka, Keiji Suzuki, Masahito Yamamoto, Azuma Ohuchi
    Proceedings of the IEEE International Conference on Systems, Man and Cybernetics 2 1999年 
    One of the important behaviors of agents in agent-based simulations is a negotiation. We focus on the price negotiations where both players have opposing interests as one of the negotiations. In the conventional price negotiation model, the process consists of an alternate succession of offers and counter-offers with exchanging the desired prices directly. As extended price negotiation model, we propose the price negotiation model whose process consists of an alternate succession of offers of desired price and counter-offers of a virtual word. We propose the agents with a unified learning system who can acquire to distinguish words and use them to negotiate. As a result, we show that the proposed agents can reach an agreement in negotiations and they acquire the negotiation strategy in evolutionary process.

MISC

  • CHANG Ming, 飯塚博幸, 成瀬康, 古川正紘, 古川正紘, 安藤英由樹, 安藤英由樹, 前田太郎, 前田太郎 ヒューマンインタフェースシンポジウム論文集(CD-ROM) 2017 2017年
  • 常明, 飯塚博幸, 成瀬康, 古川正紘, 安藤英由樹, 前田太郎 ヒューマンインタフェース学会誌 19 (1) 2017年
  • 今井 倫太, 小野 哲雄, 篠沢 一彦, 大澤 博隆, 飯塚 博幸, 硯川 潤, Michita Imai, Tetsuo Ono, Kazuhiko Shinozawa, Hirotaka Osawa, Hiroyuki Iizuka, Jun Suzurikawa 人工知能 = journal of the Japanese Society for Artificial Intelligence 31 (1) 43 -49 2016年01月01日
  • 小島 大樹, 岡 瑞起, 飯塚 博幸, 池上 高志 日本物理学会講演概要集 71 3088 -3088 2016年
  • 永田紘也, 飯塚博幸, 山本雅人, 鈴木恵二, 川村秀憲 電気・情報関係学会北海道支部連合大会講演論文集(CD-ROM) 2016 2016年
  • 永田紘也, 飯塚博幸, 山本雅人, 鈴木恵二, 川村秀憲 情報処理学会研究報告(Web) 2016 (ICS-184) 2016年
  • 横山想一郎, 飯塚博幸, 山本雅人 計測自動制御学会システムインテグレーション部門講演会(CD-ROM) 17th 2016年
  • 斎藤渉, 横山想一郎, 飯塚博幸, 山本雅人 計測自動制御学会システムインテグレーション部門講演会(CD-ROM) 17th 2016年
  • 加藤 修, 飯塚 博幸, 山本 雅人 情報科学技術フォーラム講演論文集 14 (2) 313 -314 2015年08月24日
  • 利根川 凛, 飯塚 博幸, 山本 雅人, 古川 正志, 大内 東 情報科学技術フォーラム講演論文集 14 (4) 479 -480 2015年08月24日
  • 内野 敬介, 飯塚 博幸, 山本 雅人 ロボティクス・メカトロニクス講演会講演概要集 2015 "1P1 -G08(1)"-"1P1-G08(4)" 2015年05月17日 
    Rubber Hand Illusion (RHI) is a phenomenon that a subject feels that a rubber hand (dummy hand) is their own when the rubber hand and the real hand are stroked synchronously and the subject receives the consistent sensory inputs between visual and tactile sensations. The consistent visuo-tactile sensation is a necessary condition for RHI, however we report that RHI can be induced by an asynchronous stimulations between visual and tactile sensations. In our experiments, the subject is required to cross their arms and given successive stimulations on their each hand. Our experiment is conducted in virtual environment. It is known that such stimulations cause the mis-temporal order judgment. Our results support that arm-crossing confuses the body image and the asynchronous stimulations can induce RHI.
  • 齋藤 宗太郎, 飯塚 博幸, 山本 雅人 ロボティクス・メカトロニクス講演会講演概要集 2015 "2A1 -M01(1)"-"2A1-M01(4)" 2015年05月17日 
    Perceptual Crossing experiment has been proposed to analyse the human social interaction with low dimensional sensation and motion. The previous study shows that human subjects can distinguish the partner with the distracted moving objects by coordinated behaviors established in two. In order to investigate the coordinated behavior established by human subjects, we design different types of AI programs and test how those programs can interact with human subjects. Our result shows that the turn-taking type AI program is better to reproduce human social interaction.
  • 加藤 修, 飯塚 博幸, 山本 雅人 知識ベースシステム研究会 104 7 -12 2015年03月01日
  • 南 賢一, 飯塚 博幸, 山本 雅人 インテリジェントシステム・シンポジウム講演論文集 2014 (24) 152 -155 2014年09月18日
  • 家登 亮多, 飯塚 博幸, 山本 雅人 インテリジェントシステム・シンポジウム講演論文集 2014 (24) 100 -103 2014年09月18日
  • 松林 涼平, 飯塚 博幸, 山本 雅人 インテリジェントシステム・シンポジウム講演論文集 2014 (24) 210 -213 2014年09月18日
  • 常 明, 飯塚 博幸, 成瀬 康 日本バーチャルリアリティ学会大会論文集 Proceedings of the Virtual Reality Society of Japan, Annual Conference 19 480 -483 2014年09月17日 [査読無し][通常論文]
  • 田村 康将, 飯塚 博幸, 山本 雅人 日本オペレーションズ・リサーチ学会秋季研究発表会アブストラクト集 2014 66 -67 2014年08月28日
  • 横山 想一郎, 飯塚 博幸, 山本 雅人 日本オペレーションズ・リサーチ学会秋季研究発表会アブストラクト集 2014 68 -69 2014年08月28日
  • 飯塚 博幸, 齋藤 健太郎, 山本 雅人 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報 114 (85) 17 -22 2014年06月16日 
    ロボットの構造が複雑になるにつれ,それを操縦するインタフェースも複雑なものとなる.本稿では,簡易なインタフェースにも関わらず,複雑な構造をもつたロボットを操縦するための共創造型インタフェースを提案する.ユーザである人間と機械の双方が学習することによって,人の学習負荷を軽減してより良いインタフェースを協働で作り上げる.提案手法を用いて構築された共創造型インタフェースを,多自由度なロボットの操縦において操縦性能の観点から評価した.最後に,人間と機械の共創造的インタフェースについて議論する.
  • 田村 康将, 飯塚 博幸, 山本 雅人 ロボティクス・メカトロニクス講演会講演概要集 2014 "1A1 -L03(1)"-"1A1-L03(4)" 2014年05月24日 
    The job-shop scheduling problem (JSP) is one of the combinatorial optimizations and one of the most difficult order decision problems. Many approximate algorithms, which are heuristic methods or metaheuristics, are studied to solve JSP, because JSP is categorized into NP-hard problems. The application of the dispatching rules is known as a practical heuristic for JSP or many other scheduling problems. This study proposes an extended algorithm for JSP based on the genetic algorithm and a novel local search method using the dynamic dispatching rules. The extended algorithm searches for effective dispatching rules along with schedules according to the genetic algorithm. Also the schedules are partially modified by the rules in every generation. In order to verify the effectiveness of the proposed algorithm, this paper shows the result of numerical experiments.
  • 小川 純, 飯塚 博幸, 山本 雅人, 古川 正志 ロボティクス・メカトロニクス講演会講演概要集 2014 "1A1 -K06(1)"-"1A1-K06(4)" 2014年05月24日 
    Twist of seaweeds is one of the physical problems that disturb the growth of the seaweeds in the marine biomass cultivation. This study aims to evaluate the twist of seaweeds in a quantitative way. There is no computational model to simulate dynamics of the twist of seaweeds. In this study, we create the physical model of the seaweeds using physics simulation and the states between seaweed individuals are characterized with feature vectors, which is composed the physical, geometrical, and time factors. The measure of twist can be defined as the number of features and classifies a state between individuals into "twist state" or "non-twist state".
  • 高田 圭, 本庄 将也, 飯塚 博幸, 山本 雅人 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報 113 (459) 7 -12 2014年03月03日 
    Hexは簡単なルールの二人用ボードゲームである.交互に石を置くことでゲームは進行し,先に自色の対辺を自分の色の石で接続すれば勝利となる.Hexの任意の局面は,2つの局面ネットワークで表現できることが知られている.既存研究で局面ネットワークの平均媒介中心性の重要性は示されているが,十分ではない,本研究では,熟練者同士の対戦から,盤の両端間の経路の多様性も考慮することが,重要である可能性を述べる.そこで,平均媒介中心性と盤の両端間の経路を考慮した局面評価関数を開発,検証することで,盤の両端間の経路の多様性が重要であることを明らかにする.
  • 横山想一郎, 飯塚博幸, 山本雅人 日本オペレーションズ・リサーチ学会秋季研究発表会アブストラクト集 2014 2014年
  • 横山想一郎, 飯塚博幸, 山本雅人 日本機械学会ロボティクス・メカトロニクス講演会講演論文集(CD-ROM) 2014 2014年
  • Yuki Sato, Hiroyuki Iizuka, Takashi Ikegami CONSTRUCTIVIST FOUNDATIONS 9 (1) 89 -90 2013年11月
  • 飯塚 博幸, 宇野 良子 人工知能学会誌 28 (6) 950 -950 2013年11月01日 [査読無し][通常論文]
  • 近藤 大祐, 飯塚 博幸, 安藤 英由樹 日本バーチャルリアリティ学会大会論文集 Proceedings of the Virtual Reality Society of Japan, Annual Conference 18 285 -288 2013年09月18日
  • 北尾 太嗣, 近藤 大祐, 飯塚 博幸 日本バーチャルリアリティ学会大会論文集 Proceedings of the Virtual Reality Society of Japan, Annual Conference 18 347 -350 2013年09月18日
  • Tom Froese, Hiroyuki Iizuka, Takashi Ikegami BEHAVIORAL AND BRAIN SCIENCES 36 (4) 420 -421 2013年08月 [査読無し][通常論文]
     
    Synthetic approaches to social interaction support the development of a second-person neuroscience. Agent-based models and psychological experiments can be related in a mutually informing manner. Models have the advantage of making the nonlinear brain-body-environment-body-brain system as a whole accessible to analysis by dynamical systems theory. We highlight some general principles of how social interaction can partially constitute an individual's behavior.
  • 安藤 英由樹, 近藤 大祐, 飯塚 博幸, 小濱 和貴, 坂井 義治, 前田 太郎 電子情報通信学会技術研究報告. HCS, ヒューマンコミュニケーション基礎 113 (72) 195 -200 2013年05月23日 [査読無し][通常論文]
     
    腹腔鏡下手術とは,腹腔にカメラと鉗子等を刺入して行う外科手術であり,開腹手術と比べて低侵襲性であるが,難易度が高く熟練医が不足していることが大きな問題である.従来のトレーニングツールはボックストレーナやシミュレータなどあるものの,模擬環境での繰り返し学習が中心で,どのように熟練者のスキルを学習させるかについては考慮されておらず,長期の学習時間を要する,悪い癖がつくなどの問題を抱えていた.そこで,我々は内視鏡外科手術は術野像がモニタによって同じ視野が「共有される」という特性を利用して,段階的な学習ステップに応じた,熟練者の適切かつスムースな動きを修練者の間で共有することで,非言語的なスキルのコミュニケーションを図り,効果的なトレーニングを行う手法を提案し,また効果的な評価方法について議論する.
  • 安藤 英由樹, 近藤 大祐, 飯塚 博幸, 小濱 和貴, 坂井 義治, 前田 太郎 電子情報通信学会技術研究報告. HIP, ヒューマン情報処理 113 (73) 195 -200 2013年05月23日 [査読無し][通常論文]
     
    腹腔鏡下手術とは,腹腔にカメラと鉗子等を刺入して行う外科手術であり,開腹手術と比べて低侵襲性であるが,難易度が高く熟練医が不足していることが大きな問題である.従来のトレーニングツールはボックストレーナやシミュレータなどあるものの,模擬環境での繰り返し学習が中心で,どのように熟練者のスキルを学習させるかについては考慮されておらず,長期の学習時間を要する,悪い癖がつくなどの問題を抱えていた.そこで,我々は内視鏡外科手術は術野像がモニタによって同じ視野が「共有される」という特性を利用して,段階的な学習ステップに応じた,熟練者の適切かつスムースな動きを修練者の間で共有することで,非言語的なスキルのコミュニケーションを図り,効果的なトレーニングを行う手法を提案し,また効果的な評価方法について議論する.
  • 近藤 大祐, 飯塚 博幸, 安藤 英由樹, 小濱 和貴, 坂井 義治, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2013 "1A2 -C05(1)"-"1A2-C05(4)" 2013年05月22日 
    We are developing laparoscopy training system, by adding a video synthesize functions on the conventional training box. The video of the expert's motion is shown on the monitor and is overwrapped onto the user's own view. the user can trace the expert's motion with one's forceps to learn ideal procedures. Our proposed method was evaluated in terms of learning effectiveness.
  • 飯塚 博幸, 下山 冬馬, 丹羽 真隆, 安藤 英由樹, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2013 "2A1 -F04(1)"-"2A1-F04(4)" 2013年05月22日 
    This paper investigates how to extract and transmit behavioral intentions of users from intuitive inputs of Tsumori controller. We performed two experiments. The first experiment clarified how much the robot can be delayed to start moving toward operator's input without losing operator's agency, the feeling that they are initiating the motion. The second experiment compared fundamental two real-time algorithms to select robot segmented motion and showed that the average method could be better to stabilize the operator's intuitive input. Our results are discussed in terms of agency and real-time operation for Tsumori controller.
  • 米村 朋子, 横坂 拓巳, 近藤 大祐, 橋本 悠希, 飯塚 博幸, 安藤 英由樹, 前田 太郎 基礎心理学研究 31 (2) 222 -222 2013年03月30日 [査読無し][通常論文]
  • 安藤 英由樹, 飯塚 博幸 人工知能学会誌 28 (2) 310 -317 2013年03月01日 [査読無し][通常論文]
  • 平原誠一郎, 飯塚博幸, 前田太郎, 渡邊淳司, 安藤英由樹 日本バーチャルリアリティ学会大会論文集(CD-ROM) 18th 2013年
  • 鳥居航, 近藤大祐, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 18th 2013年
  • 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会研究報告 18 (TX01) 2013年
  • 飯塚 博幸, 安藤 英由樹, 前田 太郎 人工知能学会全国大会論文集 27 1 -3 2013年
  • 宇野 良子, 飯塚 博幸, 鈴木 啓介 電子情報通信学会技術研究報告. HCS, ヒューマンコミュニケーション基礎 112 (176) 55 -60 2012年08月11日 [査読無し][通常論文]
     
    本稿は,リーダー・フォロワーの分化を,私たちが行った二つの実験から考える.一つ目は計算機シミュレーションによる実験で,エージェント同士が「主体性」の確認のために役割分化のある運動が生まれる.二つ目は,人間の被験者のペアが新しい「言語」を用いてコミュニケーションを構造化するのを観察する実験である.コミュニケーションのトピックを予め与えていないため,被験者はトピックを模索するが,その際に役割分化が現れやすい.本稿では、言語コミュニケーションにおいて主体性の確認は、トピックの模索のような意味生成を通じて行われているのではないかと論じ,二つの実験から得たモデルに基づいて様々なコミュニケーションを分析する可能性を探る.
  • 米村 朋子, 岡本 信, 近藤 大祐, 橋本 悠希, 飯塚 博幸, 安藤 英由樹, 前田 太郎 基礎心理学研究 30 (2) 234 -234 2012年03月31日 [査読無し][通常論文]
  • 横坂 拓巳, 米村 朋子, 近藤 大祐, 橋本 悠希, 飯塚 博幸, 安藤 英由樹, 前田 太郎 基礎心理学研究 30 (2) 234 -234 2012年03月31日 [査読無し][通常論文]
  • 飯塚 博幸, 安藤 英由樹, 前田 太郎 電子情報通信学会技術研究報告. HIP, ヒューマン情報処理 111 (432) 15 -18 2012年02月02日 [査読無し][通常論文]
     
    本稿では自己と他者の違いに焦点を当てる.動作を行っている主体が自分であるという感覚を運動主体感と呼ぶが,この運動主体感が視覚と触覚と運動の感覚の統合においてどのように形成されているのかを明らかにすることを目的としている.この問題を扱うに当たり,他者にくすぐられるとくすぐったいが,自分で自分をくすぐることはできないことを利用する.実験では,視覚刺激を操作することで自分でくすぐっているにも関わらず,くすぐったくなることを示し,運動主体感について議論する.
  • 近藤大祐, 近藤大祐, 米村朋子, 米村朋子, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 17th 2012年
  • 北尾太嗣, 米村朋子, 米村朋子, 横坂拓巳, 近藤大祐, 近藤大祐, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 17th 2012年
  • 青山一真, 橋本悠希, 近藤大祐, 米村朋子, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 17th 2012年
  • 橋本悠希, 橋本悠希, 近藤大祐, 米村朋子, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 17th 2012年
  • 濱口英典, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会研究報告 17 (HDC09) 2012年
  • 丹羽 真隆, 岡田 慎矢, 東 健太郎, 飯塚 博幸, 安藤 英由樹, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2011 "1P1 -H11(1)"-"1P1-H11(4)" 2011年05月26日 
    Exploiting the time-segmentation of action recognition and generation, we have proposed "Tumori" controller that realizes to manipulate a high-DOF (degree of freedom) robot with a low-DOF joystick controller while maintaining the user's intention depending on the time-segmentation. In this paper, the relation between behavior intention and two type of input device("Tumori" controller) is investigated.
  • 丹羽 真隆, 岡田 慎矢, 東 健太郎, 飯塚 博幸, 安藤 英由樹, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2011 "1P1 -H10(1)"-"1P1-H10(3)" 2011年05月26日 
    We propose novel architecture to realize a low-degree of freedom (DOF) Tsumori controller to manipulate a high-DOF robot. We believe that humans can immerse themselves to manipulate a robot if the robot reflects user intention. In our Tsumori controller, the user intuitively operates a joystick, and the robot moves semi -autonomously and reflects the user's intention that is extracted from intuitive input sequences. In this paper, we explain how to extract Tsumori, which is an archetype of human behavioral intention, and how to realize a Tsumori controller.
  • 橋本悠希, 米村朋子, 近藤大祐, 飯塚博幸, 安藤英由樹, 前田太郎 横幹連合コンファレンス(CD-ROM) 4th 2011年
  • 熊倉祥人, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 16th 2011年
  • 浜口英典, 丹羽真隆, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 16th 2011年
  • 橋本悠希, 橋本悠希, 米村朋子, 米村朋子, 飯塚博幸, 飯塚博幸, 前田太郎, 前田太郎, 安藤英由樹, 安藤英由樹, 梶本裕之, 梶本裕之 日本感性工学会大会予稿集(CD-ROM) 13th 2011年
  • 米村朋子, 米村朋子, 横坂拓巳, 近藤大祐, 橋本悠希, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 基礎心理学研究 29 (2) 2011年
  • 安藤 英由樹, 飯塚 博幸, 米村 朋子, 前田 太郎 電子情報通信学会技術研究報告. HCS, ヒューマンコミュニケーション基礎 110 (33) 137 -142 2010年05月06日 [査読無し][通常論文]
     
    本稿では,パラサイト・ヒューマン・ネットワークを用いた五感情報通信によって安全安心応用を実現するために,光学共役な視覚共有システムを実現し,このシステムによって実時間支援,追体験学習の実現可能性を示す実験とその結果からパラメータの設計指針について報告する.
  • 安藤 英由樹, 飯塚 博幸, 米村 朋子, 前田 太郎 電子情報通信学会技術研究報告. HIP, ヒューマン情報処理 110 (34) 137 -142 2010年05月06日 [査読無し][通常論文]
     
    本稿では,パラサイト・ヒューマン・ネットワークを用いた五感情報通信によって安全安心応用を実現するために,光学共役な視覚共有システムを実現し,このシステムによって実時間支援,追体験学習の実現可能性を示す実験とその結果からパラメータの設計指針について報告する.
  • 湯村武士, 橋本悠希, 橋本悠希, 米村朋子, 米村朋子, 飯塚博幸, 飯塚博幸, 安藤英由樹, 安藤英由樹, 前田太郎, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 黒崎敬太郎, 川崎宏記, 近藤大祐, 米村朋子, 米村朋子, 橋本悠希, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 熊倉祥人, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 横坂拓巳, 米村朋子, 米村朋子, 岡本信, 川崎宏記, 近藤大祐, 橋本悠希, 橋本悠希, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 岡田慎矢, 坂口翔太, 丹羽真隆, 東健太郎, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 服部響, 黒崎敬太郎, 川崎宏記, 近藤大祐, 橋本悠希, 橋本悠希, 米村朋子, 米村朋子, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 橋本悠希, 雨宮智浩, 米村朋子, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 15th 2010年
  • 近藤 大祐, 黒崎 敬太郎, 米村 朋子, 飯塚 博幸, 安藤 英由樹, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2010 "1A1 -G14(1)"-"1A1-G14(3)" 2010年 
    We are developing a remote cooperative work system using a view sharing. The view sharing is constructed from the video-see-through head mounted displays (VST-HMD) and motion trackers. Using the system, the users can share the first-person views each other. The image stabilization method is implemented to correspond the optical flow to head movement.
  • 川崎 宏記, 岡本 信, 飯塚 博幸, 安藤 英由樹, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2010 "1A1 -G13(1)"-"1A1-G13(4)" 2010年 
    When two distant persons establish coordination in the tele-existence environment, it is important how to share each other's motions and sensations, and how to adjust or revise their motions to cooperate. There are some studies in which they use a third-person view to share motions, and less studies with first-person view despite the fact that there are reports where first-person view plays an important role for ownership and bodyimage. Therefore, we propose a method to blend the mutual first person's views to establish coordination where the partner's view is superimposed onto his/her own view. In the blended views, each other's motion is seen in an equivalent viewpoint. In the experiment, subjects are required to establish rhythmic coordination under this view-sharing system, and we evaluate a performance for the rhythmic coordination in different composition-ratio conditions of the first-person views.
  • 濱口 英典, 安藤 英由樹, 飯塚 博幸, 前田 太郎 電子情報通信学会技術研究報告. HIP, ヒューマン情報処理 109 (83) 97 -100 2009年06月08日 [査読無し][通常論文]
     
    人の非線形な知覚特性を考慮し,偏加速度を持ったオモリの運動を利用した擬似牽引力の提示デバイスが研究されている.しかし,従来手法ではクランクスライダ機構のため小型・軽量化が困難であった.そこで,本研究では小型軽量で従来と同等の偏加速度運動を実現する機構として回転する2節リンクの先端にオモリを載せ,固定された壁面に周期的に衝突させる機構を提案する.本装置では衝突時の打撃力によって短時間の大きな推力を生成し,回転時のオモリの可動によって長時間の小さな推力を生成する.本発表では,効果的な偏加速度運動を生成するために,衝突部の構造,回転時のオモリの軌道,の2点に関する設計論について述べる.
  • 安藤 英由樹, 湯村 武士, 飯塚 博幸, 渡邊 淳司, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2009 "2A2 -K01(1)"-"2A2-K01(4)" 2009年05月25日 
    In this paper, we propose a method to transmit tactile information from one person to another using nail-mounted tactile display. Our proposed method conveys tactile information which is obtained by detecting the vibration on the nail of a sender when tracing a texture, and generates a vibration on the nail of a receiver to re-create the tactile perception while the subject tracing on a flat surface. Therefore, we measured the vibration on the subject's nail when tracing different kinds of textures, and analyzed its characteristics to find essential factors in order to share tactile perceptions with two people.
  • 飯塚 博幸, 安藤 英由樹, 前田 太郎 ロボティクス・メカトロニクス講演会講演概要集 2009 "1P1 -E09(1)"-"1P1-E09(4)" 2009年05月25日 [査読無し][通常論文]
     
    To establish a harmonic interaction between a human and a computer, we reconsider human interactions that establish natural communication. Instead of investigating verbal communication, we use the simple and low-level environment where inputs and outputs of subjects are restricted in a low dimension at the behavioral level. Our final goal is to clarify with the simple environment how humans can create human-like interaction such as an interaction that produces the feeling of human. To achieve it, behavior-based turing-test is proposed here. As a result, it is shown that the emergence of turn-taking behavior plays an important role to pass our behavior-based turing test. The results are also discussed with the attractor superimposition.
  • 山下 晃弘, 川村 秀憲, 飯塚 博幸, 大内 東 情報処理学会論文誌数理モデル化と応用(TOM) 2 (2) 47 -57 2009年03月27日 [査読無し][通常論文]
     
    Web 上の情報量拡大にともなって,オンラインストアなどでは推薦システムを導入する事例が増加している.同時に,これまで様々な推薦アルゴリズムが提案,分析,評価されてきた.システムによる推薦結果は,次にユーザが閲覧するアイテムとなり,結果としてどのアイテムが次に評価されるかに影響を及ぼす.したがって,推薦システムとユーザは相互作用を及ぼしあう関係にあるといえる.しかし,先行研究による推薦アルゴリズムの評価・分析では,一時的な推薦精度のみに着目し,このような相互作用は考慮されてこなかった.そこで本研究では,ユーザをエージェントとしてモデル化することで,推薦システムとユーザの相互作用を考慮した,新たな評価・分析モデルを提案する.さらに,本論文では,基本的な推薦アルゴリズムである 「ユーザ間協調フィルタリング」 と,実システムとしてすでに実用化されている 「アイテム間協調フィルタリング」 を例として取り上げ,その特性について提案モデルを用いて分析した.その結果,既存の評価・分析モデルの枠を超え,より有効な推薦への指針を検証可能であることが示された.The problem of information overload spreading across the Internet has been causing serious inefficiency in browsing and searching for information. As a way to overcome the problem, the recommender systems are recently used in many E-commerce sites. Many algorithms have been proposed to improve the accuracy of recommendation based on user ratings. The relation between recommender systems and users is rather interactive in the sense that recommendations decides which items are recommended to users and the results of ratings by users will affect the next recommendations. However, conventional studies have not considered the interactive aspects so much. Therefore, our aim of this paper is to propose a new evaluation model using multiagent modeling where the recommender system and agents (as users) interacts with each other. The properties of typical recommendation algorithms such as user-based and itembased collaborative filtering will be analyzed with our proposed model. Our results also suggest the possibilities to propose a novel and effective recommendation algorithm.
  • 長江依奈, 安藤英由樹, 安藤英由樹, 米村朋子, 米村朋子, 飯塚博幸, 飯塚博幸, 前田太郎, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 14th 2009年
  • 川崎宏記, 岡本信, 中村和志, 中村和志, 米村朋子, 米村朋子, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 14th 2009年
  • 岡田慎矢, 坂口翔太, 丹羽真隆, 飯塚博幸, 安藤英由樹, 前田太郎, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 14th 2009年
  • 廣瀬光彦, 近藤大祐, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 14th 2009年
  • 黒崎敬太郎, 濱田和孝, 近藤大祐, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 14th 2009年
  • 宮田祐樹, 濱田和孝, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 14th 2009年
  • 前田 太郎, 安藤 英由樹, 飯塚 博幸 情報処理学会研究報告. CVIM, [コンピュータビジョンとイメージメディア] 2008 (115) 81 -90 2008年11月20日 [査読無し][通常論文]
     
    大阪大学大学院情報科学研究科が取り組むグローバルCOEプログラム「アンビエント情報社会基盤創成拠点」においては生物の原理に学び,環境と人間がインタラクションし,状況に応じて柔軟に適応可能なアンビエントインタフェースの実現を目指している.本報告ではこのようなインタフェース技術として,環境側に実装する従来の環境知能型のインタフェースではなく,人間側に装着することで機能するウェアラブルインタフェース,パラサイトヒューマンについて提唱し,人間と協調して機能するアトラクター重畳概念による人間行動支援について解説する.
  • 前田 太郎, 安藤 英由樹, 飯塚 博幸 電子情報通信学会技術研究報告. PRMU, パターン認識・メディア理解 108 (327) 81 -90 2008年11月20日 [査読無し][通常論文]
     
    大阪大学大学院情報科学研究科が取り組むグローバルCOEプログラム「アンビエント情報社会基盤創成拠点」においては生物の原理に学び,環境と人間がインタラクションし,状況に応じて柔軟に適応可能なアンビエントインタフェースの実現を目指している.本報告ではこのようなインタフェース技術として,環境側に実装する従来の環境知能型のインタフェースではなく,人間側に装着することで機能するウェアラブルインタフェース,パラサイトヒューマンについて提唱し,人間と協調して機能するアトラクター重畳概念による人間行動支援について解説する.
  • 前田 太郎, 安藤 英由樹, 飯塚 博幸 電子情報通信学会技術研究報告. MVE, マルチメディア・仮想環境基礎 108 (328) 81 -90 2008年11月20日 [査読無し][通常論文]
     
    大阪大学大学院情報科学研究科が取り組むグローバルCOEプログラム「アンビエント情報社会基盤創成拠点」においては生物の原理に学び,環境と人間がインタラクションし,状況に応じて柔軟に適応可能なアンビエントインタフェースの実現を目指している.本報告ではこのようなインタフェース技術として,環境側に実装する従来の環境知能型のインタフェースではなく,人間側に装着することで機能するウェアラブルインタフェース,パラサイトヒューマンについて提唱し,人間と協調して機能するアトラクター重畳概念による人間行動支援について解説する.
  • 飯塚 博幸 情報処理学会研究報告知能と複雑系(ICS) 2008 (20) 55 -60 2008年03月06日 
    恒常性を用いたニューラルネットワークモデルの拡張を行う.従来のモデルでは,内的な恒常性と振る舞いの安定性は必ずしも関連性を形成して進化するとは限らない.つまり,振る舞いが壊れてしまったとしても,内的な恒常性が影響を受けないことがある.提案方法では,それを静止膜電位と評価方法を改良することによって,振る舞いと恒常性が密接な関連性をもった構造が進化するように拡張する.この拡張したニューラルネットワークモデルをシミュレートされた移動体に実装し,環境変動に対する適応能力を検証する.結果として提案手法はより適応的振る舞いを示し,複数回起こる環境変動に対しても飽和することなく適応し続けることを示す.A homeostatic neural controller is extended by exploiting the internal dynamics of a neural network in the absence of sensory input and also by avoiding the loss of internal variability that is caused by a local plastic mechanism. The extended method can solve a contingent link between internal and behavoiural stability where homeostasis remain unaffected even under disruptions of behaviours. The method allows the generation of reliable adaptation to morphological disruptions in a simple simulated vehicle using a homeostatic neurocontroller that has been selected to behave homeostatically while performing the desired behaviour but non-homeostatically in other circumstances. The performance of the extended method is compared with simple homeostatic neural controllers. As a result, it is shown that the extended homeostatic networks are more adaptive and can adapt to the environmental changes that happen more than once.
  • 山下 晃弘, 川村 秀憲, 飯塚 博幸, 大内 東 情報処理学会研究報告. MPS, 数理モデル化と問題解決研究報告 68 (17) 141 -144 2008年03月04日 [査読無し][通常論文]
     
    Web上の情報量が拡大する中,推薦システムが注目されている.これまでに,ユーザの行動履歴を入力として推薦の精度を最大化する手法について多くの研究が報告されてきた.しかし,推薦結果がその後のユーザ行動に影響することも考慮すると,推薦システムとユーザは相互に影響を及ぼしあう関係であるにも関わらず,そのようなダイナミクスはこれまでほとんど議論されてこなかった.そこで本研究では,推薦システムとユーザの相互作用という観点に着目し、推薦システムの利用者を購入格付けエージェントとして全体をモデル化することで推薦アルゴリズムの特長分析を行った.本稿では,推薦アルゴリズムの原点であるユーザ間協調フィルタリングと,現在一般的に利用されているアイテム間協調フィルタリングを取り上げその結果について報告する.
  • 飯塚 博幸 電子情報通信学会技術研究報告. AI, 人工知能と知識処理 107 (523) 55 -60 2008年02月27日 
    恒常性を用いたニューラルネットワークモデルの拡張を行う.従来のモデルでは,内的な恒常性と振る舞いの安定性は必ずしも関連性を形成して進化するとは限らない.つまり,振る舞いが壊れてしまったとしても,内的な恒常性が影響を受けないことがある.提案方法では,それを静止膜電位と評価方法を改良することによって,振る舞いと恒常性が密接な関連性を持った構造が進化するように拡張する.この拡張したニューラルネットワークモデルをシミュレートされた移動体に実装し,環境変動に対する適応能力を検証する.結果として提案手法はより適応的振る舞いを示し,複数回起こる環境変動に対しても飽和することなく適応し続けることを示す.
  • 麻生紘己, 岡本信, 川崎宏記, 飯塚博幸, 安藤英由樹, 前田太郎 日本バーチャルリアリティ学会大会論文集(CD-ROM) 13th 2008年
  • IIZUKA Hiroyuki, SUZUKI Keiji, YAMAMOTO Masahito, OHUCHI Azuma IEICE transactions on fundamentals of electronics, communications and computer sciences 83 (6) 1075 -1082 2000年06月25日 
    Agent-based simulations are expected to enable analysis of complex social phenomena. In such simulations, one of the important behaviors of the agents is negotiation. Through-out the negotiations, the agents can make complex interactions with each other. Therefore, the ability of agents to perform negotiation is important in simulations of artificial societies. In this paper, we focus on price negotiations, in which the two sides have opposing interests. In the conventional price negotiation model, the process consists of an alternate succession of directly presented offers and counter-offers exchanging the desired prices. As an extended price negotiation model, we introduce virtual words to mimic the negotiation techniques of humans for indirectly presenting the desired price. The process of the proposed negotiation model consists of an alternate succession of offers of desired price and counter-offers of a word. The words represent the degree of the agent's demand. We propose agents with reinforcement learning who can acquire the ability to distinguish words and use them to negotiate. As a result, we will show that the virtual words became meaningful in the process of negotiations between agents whose negotiating strategies are acquired by reinforcement leaning.

講演・口頭発表等

  • Complexity Growth by Cooperative Interaction in Adversarial Learning  [通常講演]
    Taiki Sasaki, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Twenty-sixth international symposium on Artificial life and Robotics 2021年01月 口頭発表(一般)
  • Multi-agent formation by learning communication signals using DIAL in a distributed multi-agent environment  [通常講演]
    Hironobu Horiuchi, Wataru Noguchi, Iizuka Hiroyuki, Masahito Yamamoto
    Twenty-sixth international symposium on Artificial life and Robotics 2021年01月 口頭発表(一般)
  • 周辺地域の予測レーダ雨量を用いた深層学習による水路の水位予測  [通常講演]
    石川拓実, 阿曽克司, 羽黒厚志, 野口渉, 飯塚博幸, 山本雅人
    第21回計測自動制御学会システムインテグレーション部門講演会 2020年12月 口頭発表(一般)
  • カメラ映像を用いたYOLOとI3Dによる飼育動物の行動分類  [通常講演]
    加藤圭吾, 野口渉, 飯塚博幸, 山本雅人
    第21回計測自動制御学会システムインテグレーション部門講演会 2020年12月 口頭発表(一般)
  • Semi-supervised Learning Combining 2DCNNs and Video Compression for Action Recognition  [通常講演]
    Hayato Terao, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    2020 International Conference on Vision, Image and Signal Processing(ICVISP 2020) 2020年12月 口頭発表(一般)
  • DIALを用いた通信シグナルの学習によるマルチエージェントフォーメーション  [通常講演]
    堀内宏信, 野口渉, 飯塚博幸, 山本雅人
    19回情報科学技術フォーラム(FIT2020)講演論文集, 第2分冊 2020年09月 口頭発表(一般)
  • 複数の代表点を用いたSDMMに基づく半教師あり行動認識手法  [通常講演]
    寺尾 颯人, 野口 渉, 飯塚 博幸, 山本 雅人
    19回情報科学技術フォーラム(FIT2020) 2020年09月 口頭発表(一般)
  • Learning of Evaluation Function in Digital Curling Considering the Probability of Scores at Each End  [通常講演]
    Kotaro Ataka, Wataru Noguchi, Iizuka Hiroyuki, Masahito Yamamoto
    AAAI20 Workshop, Artificial Intelligence in Team Sports 2020年02月 口頭発表(一般)
  • MixMatch based semi-supervised learning for video action recognition using 3D convolutional neural networks  [通常講演]
    Hayato Terao, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    Twenty-fifth international symposium on Artificial life and Robotics 2020年01月 口頭発表(一般)
  • Efffect of Reactivity Artificial Medaka on Real Medaka  [通常講演]
    Katsuki Auchi, Wataru Noguchi, Hiroyuki Iizuka, Masahito Yamamoto
    twenty-fifth international symposium on Artificial life and Robotics 2020年01月 口頭発表(一般)
  • Reinforcement Learning Method with Internal World Model Training  [通常講演]
    Kenji Hirata, Hiroyuki Iizuka, Masahito Yamamoto
    2020 IEEE/SICE International Symposium on System Integration 2020年01月 口頭発表(一般)
  • Analysis of Time Series Trained by Adversarial Imitation Learning in Discrete State Neural Network Model  [通常講演]
    Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    2020 IEEE/SICE International Symposium on System Integration 2020年01月 口頭発表(一般)
  • Analysis of Time Series Generated By Long Short-Term Memory Trained with Adversarial Imitation Learning[Seiya Yamazaki,Hiroyuki Iizuka,Masahito Yamamoto]  [通常講演]
    Seiya Yamazaki, Hiroyuki Iizuka, Masahito Yamamoto
    2019 IEEE Symposium Series on Computational Intelligence (IEEE SSCI 2019) 2019年12月 口頭発表(一般)
  • 分割した集団の学習による役割分担の獲得  [通常講演]
    荒木関 渉, 野口 渉, 飯塚 博幸, 山本 雅人
    情報処理北海道シンポジウム2019 2019年10月 ポスター発表
  • Hybrid Synthetic Approach to Animal Interaction  [通常講演]
    Hiroyuki Iizuka, Yosuke Nakamoto, Masahito Yamamoto
    The 2019 Conference on Artificial Life 2019年07月 口頭発表(一般)
  • Spatial Representation of Self and Other by Superposition Neural Network Model  [通常講演]
    Wataru Noguchi, Hiroyuki Iizuka, Shigeru Taguchi, Masahito Yamamoto
    The 2019 Conference on Artificial Life 2019年07月 口頭発表(一般)
  • 3次元畳み込みニューラルネットワークを用いた店舗における顧客行動の分類  [通常講演]
    貝塚 裕, 飯塚 博幸, 山本 雅人
    ロボティクス・メカトロニクス講演会2019 (ROBOMECH2019) 2019年06月 ポスター発表

共同研究・競争的資金等の研究課題

  • 日本学術振興会:科学研究費助成事業
    研究期間 : 2022年04月 -2026年03月 
    代表者 : 飯塚 博幸, 山本 雅人, 鈴木 啓介, 野口 渉
  • 日本学術振興会:科学研究費助成事業
    研究期間 : 2022年04月 -2025年03月 
    代表者 : 山本 雅人, 飯塚 博幸, 田村 康将, 野口 渉
  • 日本学術振興会:科学研究費助成事業 新学術領域研究(研究領域提案型)
    研究期間 : 2020年04月 -2022年03月 
    代表者 : 飯塚 博幸
     
    本研究では,認知実験とコンピュータによる構成論的アプローチによって,単純なシグナルのやりとりでの「真似したいが真似されたくない」という敵対的模倣関係のなかから,そのシグナル上にいかにして言語的な階層構造が意図共有を通して創発可能かを明らかにすることである.初年度においては,Kirbyらの行った累積的文化進化の実験室実験を拡張することで実験する予定であったが,対面での実験が難しかったため,実験装置の作成と小規模な予備実験に留めた.実験ではサイモンゲームを用いて,味方プレイヤには真似されるように,敵プレイヤには真似されないように色パターンを生成する.予備実験結果においては,相手には真似されないように色パターン系列を生成すると,生成されるパターンの複雑化がシミュレーションと同様に見られることを明らかにした.生成されたパターン系列は,複雑さの指標として圧縮率と構造化の指標としてSQUITURアルゴリズムを用いて評価した.しかし,生成されるパターンにおいて,構造化されるようなものは得られなかった.一方で,シミュレーションでは,機械学習で注目を集めている敵対的生成ネットワークを利用して,敵対的模倣学習のモデル化を行った.敵対的生成ネットワークでは,実画像を利用することで,実画像と同程度の複雑さをもち,かつ,リアリティを人に感じさせる構造的特徴を学習可能である.これを利用して敵対的模倣学習をモデル化することで,前回の公募課題のときには見られなかった2者間において,時系列パターンの複雑が生じることを示した.
  • 日本学術振興会:科学研究費助成事業 新学術領域研究(研究領域提案型)
    研究期間 : 2018年04月 -2020年03月 
    代表者 : 飯塚 博幸
     
    初年度において関数の表現が豊かなフィードフォワードニューラルネットワークを用いても,相手を真似したいが,相手からは真似されたくない状況である敵対的模倣学習によって時系列が複雑化することが明らかになった.最終年度は,まず,敵対的模倣が複雑化に寄与しているのかを確かめるために,異なる真似関係構造がもたらす時系列の複雑さを明らかにした.相互作用の種類として,2体のそれぞれのエージェントに,模倣のみ(真似したい),被模倣のみ(真似されたくない),敵対的模倣の3種類を用いてシミュレーションを行った.結果として,少なくとも一方が敵対的模倣のときのみ時系列は複雑化し,そのときに相手が被模倣もしくは敵対的である必要があることが明らかとなった.両者の真似されたくないというのが複雑化の必要条件であり,かつ,少なくともどちらか一方が真似したいという敵対的模倣が必要であるとわかった. さらに,敵対的模倣学習がもたらす時系列の時間方向の複雑化を示すために,記憶を保持できるリカレントニューラルネットワークの一つであるLSTMを用いてモデルの構築とシミュレーションを行った.結果,初年度のモデルと同様に,リアプノフ指数が正となり,時系列が複雑化することがわかった.そこで,記憶のないモデル時に見られなかったような時間方向への構造化を定量的に示すために,ZIP圧縮を用いて生成時系列を圧縮させた.時系列がカオスになっているエポックのうち,その中でも相対的に圧縮できるパターンが生まれていることが明らかとなった.このときの時系列の遷移パターンを可視化するために,N-gramモデルを用いて時系列を有限状態文法で表し,ノード数・パス数ともに小さくなっていることを示し,より単純なグラフになっていることがわかった.記憶を持つ場合には,空間方向だけでなく時間方向も扱えるため,時間方向の複雑化・構造化が生じた.
  • 日本学術振興会:科学研究費助成事業 基盤研究(C)
    研究期間 : 2017年04月 -2020年03月 
    代表者 : 飯塚 博幸, 山本 雅人
     
    メダカを実験の対象とし,リアルタイムに検出したメダカの位置に応じて視覚刺激を生成し提示する双方向相互作用システムを構築した.この実験装置を使用して,人工物とメダカの間に双方向相互作用が可能である条件と,単方向相互作用条件とを比較して,人工物とメダカの間に双方向相互作用が成立可能かを実験した.相互作用条件のときにリアルのメダカは人工物に興味を示し,単方向相互作用実験よりも人工物の存在する画面の方に滞在する時間が増加した.つまり,双方向の時に,リアルのメダカの特定の行動を誘発することが可能であった.これはコミュニケーション様の相互作用を人工物とメダカの間に実現できたといえる.
  • 日本学術振興会:科学研究費助成事業 基盤研究(C)
    研究期間 : 2015年04月 -2018年03月 
    代表者 : 山本 雅人, 飯塚 博幸
     
    本研究では,ユーザの身体動作の情報などの時系列情報から意図を抽出することで,ユーザの意図通りに機械を操作可能とするインタフェースの開発を行った.特に,多自由度ロボットの操作を対象とし,あらかじめ決められたコマンドを覚えることなく,動作を繰り返す中で自ら学習し,それらの入力と動作の関係から機械も学習することでユーザが直感的に操作しやすいインタフェースを構築した.
  • 日本学術振興会:科学研究費助成事業 新学術領域研究(研究領域提案型)
    研究期間 : 2015年04月 -2017年03月 
    代表者 : 飯塚 博幸
     
    前年度においてPerceptual Crossing実験を行うことで,人の非言語相互作用に時間遅れをいれた場合,400msから500msの時間遅れが入ってしまうと例え人が操作していたとしても人らしさが大きく失われることがわかった.これは,ロボットが人と相互作用を行っている状況に当てはめると,相手の動作に対する人の応答タイミングから,400ms以上ロボットの動作が遅れてしまうとロボットの動きは人らしくなり得ないということを意味している.これを改善するために本年度において,引き続き人同士の相互作用の解析を行い,人がロボットを操縦する際,ロボットが人らしく振る舞うための方法論の構築を行った.まず,時間遅れに対して人は振る舞いを変えることで相互作用の頑健性を上げていることがわかった.そこでは,両者が相互作用を行うときに生じる接触を意味する振動刺激の間隔が遅くなっていることがわかった.これは両者の相互作用のリズムを遅くすることで,時間遅れが相互作用に入ったときにお互いの応答タイミングに幅が生じたためだと考えられる.次に,この人の適応行動を利用することで,ロボットの応答タイミングが遅れても人らしく振る舞うことのできる方法を提案した.アイデアとしては,外部からインタフェースとして相互作用のリズムを遅くしてやることで,相互作用の頑健性を向上できるではないかと考えた.方法としては,Perceptual Crossing実験内での運動ゲインを下げ,運動の効果を落とした.これは必ずしも相互作用のリズムを遅くすることにはつながらないが,結果的に相互作用のリズムを落とすことに成功した.これにより,相互作用の頑健性の向上も観察された.また,定量的に移動エントロピーを用いることで,時間遅れがあっても相互作用が改善されていることを示した.
  • 日本学術振興会:科学研究費助成事業 挑戦的萌芽研究
    研究期間 : 2013年04月 -2015年03月 
    代表者 : 飯塚 博幸
     
    通常,ニューロフィードバックは自分の脳波を観察することによって行われるが,本研究では,計測された脳波を直接伝送し,2者の間においてニューロフィードバックを行う実験を行った.フィードバックする脳波はα波帯域のパワーを視覚刺激で表示する方法を用いた.結果として,この相互の脳波を伝え合うコミュニケーション状態を利用し,双方のα波を高める訓練が成功した.2者間でのコミュニケーション型ニューロフィードバックで訓練をした場合には,その後,1者でニューロフィードバックを行ってもその効果が継続していることがわかった.
  • 日本学術振興会:科学研究費助成事業 基盤研究(A)
    研究期間 : 2010年 -2012年 
    代表者 : 前田 太郎, 安藤 英由樹, 飯塚 博幸, 鈴木 隆文
     
    本研究では,ヒトの運動計画の時間分節性を利用して行動意図のアーキタイプとしての「つもり」の推定とその活用について研究を展開し,多自由度の人型ロボットの操縦制御において,「つもり」の分類と推定,さらに工学的実現性において実時間性への対応を可能にしつつ随意感の発生条件を拡張的に緩和することに成功した.また脳波計測を用いて「つもり」の機能的な実在と生理的な指標の対応付けにも成功しており,脳機能のモデル化によって行動意図「つもり」の応用範囲を拡大していけるものと期待される.
  • 日本学術振興会:科学研究費助成事業 若手研究(B)
    研究期間 : 2010年 -2011年 
    代表者 : 飯塚 博幸
     
    自己と他者の違いについて運動主体間を用いることで焦点を当てる.動作を行っている主体が自分であるという運動主体感が視覚と触覚と運動の感覚の統合においてどのように形成されているのかを明らかにすることを目的とし,他者にくすぐられるとくすぐったいが,自分で自分をくすぐることはできないことを利用する.結果として視覚刺激を操作することで自分でくすぐっているにも関わらず,くすぐったくなることを示し,運動主体感の生起について明らかにした.
  • 日本学術振興会:科学研究費助成事業 特別研究員奨励費
    研究期間 : 2005年 -2007年 
    代表者 : 飯塚 博幸
     
    本年度は,実際に認知実験に見られる主体性に関する認知実験をもとにして,相互作用における能動性と主体性のモデル化を行った.人間は他人との相互作用において,自己の意図性をそのやりとりされる運動の中に表現する.その主体性は一人の運動の表現から導かれるものではなく,相互作用そのものに埋め込まれている(例えば,Trevarthenらの実験(1985)).この間主観的な社会的相互作用における主体性の検出運動ダイナミクスのシミュレーションを行い,大域的な力学系の観点から解析を行った. シミュレーションでは,相手のエージェントがニューラルネットワークを使ってリアルタイムに応答しているのか,それとも,以前相互作用したときに録画された運動をリプレイするだけの非応答エージェントであるのか,を区別することをもって相手の主体性の判断とした. この区別ができるエージェントを進化的アルゴリズムによって獲得し,その運動と内部ダイナミクスの解析を行った.結果として,ここでのエージェントの主体性はアトラクター間のトランジェントダイナミクスが相互に絡み合うことによって実現された.また,その主体性検出のための運動ダイナミクスは,ノイズに対する大域的なダイナミクスのロバスト性と接触タイミングを揺らがすための入力に対する鋭敏性と振る舞いの多様性を利用することで十分可能であることがわかった.このことは,相手の主体性を認知するために,複雑な情報を統合する個々の認知機構は必ずしも必要条件とはならず,むしろ,相互作用の自体のリアルタイム性とノイズに対する頑健性によって実現が可能であることを意味している.これらのシミュレーション結果は,人の主体性を扱うような難解な認知の問題に対しても,モデルを複雑化して理解するのではなく,環境とエージェントを簡素に記述し,その相互作用から考えるという身体性モデリングの一つの可能性を示した.
  • 日本学術振興会:科学研究費助成事業 特別研究員奨励費
    研究期間 : 2001年 -2003年 
    代表者 : 飯塚 博幸
     
    協調的でもあり競争的でもある交渉という環境において、そのお互いの目的が協調的でかつ競争的であるシミュレーション環境のモデル化を行い、そのような環境における各エージェントの学習機構の開発とそのエージェントが学習によって獲得した内部構造の解析を行った。さらに、2エージェント間の合意形成の1つとして、ここでは、turn-takeの構造に着目した。Turn-takeの振る舞いは自分の目的を達成しながらも、相手の要求に応えて自分の目的を変化させるという交捗過程には必要な構造を含んでいる。通常、一人の特定の相手に対する適応的な戦略構造等に主眼がおかれるが、ここでは、未知の相手との協調行動の創発、適応可能性について注目した。 シミュレーションの結果、幾何学的な軌跡を描くgeometric turn-taking、軌跡やターンの切り替えが規則的に生じないchaotic turn-taking、ノイズの不安定性を利用するnoise-driven turn-takingの3つのカテゴリーに分けられる協調行動としての運動が得られた。これらの進化して得られたエージェントを、世代を変えて相互作用をさせ、得られている振る舞いの複雑さと2者の協調での運動の関係について調べた。geometric turn-takingを行うエージェントは限定された相手とturn-takingを構築することができる。一方、chaotic turn-takingを行うエージェントは比較するとgeometricなものより多くのものとturn-takingを行うことが可能であることがわかった。つまり、協調行動に対して、振る舞いの多様性が、従来の特定の相手に対する適応行動だけでなく、未知の相手に対する潜在的な適応行動を保持しうることをシミュレーションによって示した。こうしたadaptabilityのシミュレーションの結果は本研究が初めてである。


Copyright © MEDIA FUSION Co.,Ltd. All rights reserved.