稲葉 通将

人工知能先端研究センター准教授
情報学専攻准教授
Ⅰ類(情報系)准教授

学位

  • 博士(情報科学), 名古屋大学

研究キーワード

  • ヒューマンコンピュータインタラクション
  • 自然言語処理
  • 人工知能
  • 対話システム

研究分野

  • 情報通信, エンタテインメント、ゲーム情報学
  • 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, 知能情報学

経歴

  • 2019年04月 - 現在
    電気通信大学, 人工知能先端研究センター, 准教授
  • 2012年04月 - 2019年03月
    広島市立大学, 大学院情報科学研究科, 助教

学歴

  • 2010年04月 - 2012年03月
    名古屋大学, 大学院情報科学研究科

委員歴

  • 2023年04月 - 現在
    専門委員, 人工知能学会 言語・音声理解と対話処理研究会, 学協会
  • 2021年05月 - 現在
    編集委員, 人工知能学会, 学協会
  • 2020年09月 - 現在
    編集委員, 言語処理学会, 学協会
  • 2020年04月 - 2023年03月
    幹事, 人工知能学会 言語・音声理解と対話処理研究会, 学協会

受賞

  • 受賞日 2021年11月
    情報処理学会
    説得されることでユーザを楽しませる人狼ゲーム対話システムの構築
    情報処理学会論文誌ジャーナル/JIP特選論文, 稲葉通将
  • 受賞日 2016年11月
    雑談ばかりしていると殺される?人狼ゲームにおける発話行為タグセットの提案とプレイヤの行動・勝敗の分析
    情報処理学会論文誌ジャーナル/JIP特選論文, 稲葉 通将;大畠 菜央実;高橋 健一;鳥海 不二夫
  • 受賞日 2016年06月
    第30回人工知能学会全国大会 (JSAI2016),
    LSTMエンコーダを用いた統計的発話選択
    全国大会優秀賞, 稲葉 通将
  • 受賞日 2014年09月
    人狼ゲームにおける役職の強さの定量化
    FANプレゼンテーション賞, 稲葉 通将

論文

  • 対話システムライブコンペティション6
    東中 竜一郎; 高橋 哲朗; 稲葉 通将; 斉 志揚; 佐々木 裕多; 船越 孝太郎; 守屋 彰二; 佐藤 志貴; 港 隆史; 境 くりま; 船山 智; 小室 允人; 西川 寛之; 牧野 遼作; 菊池 浩史; 宇佐美 まゆみ
    人工知能学会研究会資料 言語・音声理解と対話処理研究会, 一般社団法人 人工知能学会, 99巻, 掲載ページ 84-89, 出版日 2023年12月04日, 本稿では,第6回対話システムライブコンペティションの概要を述べる.本コンペティションではマルチモーダル対話システムを対象とし,特定のシチュエーションにおいて人間らしく対話を行うシチュエーショントラックについてコンペティションが実施された.本稿ではシチュエーショントラックの仕様を述べるとともに,予選の結果について述べる.
    日本語
  • Generating Character Lines in Four-Panel Manga
    Michimasa Inaba
    The 37th Pacific Asia Conference on Language, Information and Computation (PACLIC2023), 出版日 2023年12月
    研究論文(国際会議プロシーディングス)
  • SumRec: A Framework for Recommendation using Open-Domain Dialogue
    Ryutaro Asahara; Masaki Takahashi; Chiho Iwahashi; Michimasa Inaba
    The 37th Pacific Asia Conference on Language, Information and Computation (PACLIC2023), 出版日 2023年12月
    研究論文(国際会議プロシーディングス)
  • 対話状況やユーザに適応的に振る舞う対話システムのための基盤構築
    東中 竜一郎; 駒谷 和範; 宮尾 祐介; 稲葉 通将; 港 隆史; 槇原 靖
    人工知能, 38巻, 5号, 掲載ページ 714-721, 出版日 2023年09月
    研究論文(学術雑誌), 日本語
  • AIWolfDial 2023: Summary of Natural Language Division of 5th International AIWolf Contest.
    Yoshinobu Kano; Neo Watanabe; Kaito Kagaminuma; Claus Aranha; Jaewon Lee; Benedek Hauer; Hisaichi Shibata; Soichiro Miki; Yuta Nakamura; Takuya Okubo; Soga Shigemura; Rei Ito; Kazuki Takashima; Tomoki Fukuda; Masahiro Wakutani; Tomoya Hatanaka; Mami Uchida; Mikio Abe; Akihiro Mikami; Takashi Otsuki; Zhiyang Qi; Kei Harada; Michimasa Inaba; Daisuke Katagami; Hirotaka Osawa; Fujio Toriumi
    INLG (Generation Challenges), 掲載ページ 84-100, 出版日 2023年
    研究論文(国際会議プロシーディングス)
  • Collection and Analysis of Travel Agency Task Dialogues with Age-Diverse Speakers.
    Michimasa Inaba; Yuya Chiba; Ryuichiro Higashinaka; Kazunori Komatani; Yusuke Miyao; Takayuki Nagai
    LREC, 掲載ページ 5759-5767, 出版日 2022年06月
    研究論文(国際会議プロシーディングス)
  • 説得されることでユーザを楽しませる人狼ゲーム対話システムの構築
    稲葉, 通将
    情報処理学会論文誌, 62巻, 11号, 掲載ページ 1779-1791, 出版日 2021年11月15日, 本論文は,プレイヤ間で会話することで進行するパーティゲーム「人狼ゲーム」を題材とし,自然言語によりユーザに上手に説得されることでユーザを楽しませる対話システムの構築手法を提案する.提案システムはゲームの状況とユーザとの対話の流れに応じて適切な応答を複数の応答候補から選択することで対話を進めるものであり,そのためのマルチタスク学習に基づくニューラルネットワークモデルを提案する.また,自然な説得対話の実現には,システム側はユーザの説得に徐々に応じるように態度を少しずつ変化させる必要がある.そのための手法として,自然な説得の受諾を実現するためのデータドリブンな説得対話管理手法を提案する.実験ではデータに基づく客観評価とユーザとシステムが実際に対話を行う主観評価の両方を実施し,ベースラインシステムと比較して自然かつ楽しい対話が可能であることが確認された.
    This paper proposes a construction method for the persuadee dialogue system for entertainment, which is to be persuaded by users to provide them with amusement. We focuse on the dialogue in the werewolf game (also known as the mafia game) where the system simulates a persuasive dialogue. In developing the system, this study proposes a multi-task neural network for suitable response selection and a data-driven dialogue management method for naturally accepting users' persuasion.
    研究論文(学術雑誌), 日本語
  • Selection of Evaluation Metrics for Dialogue Breakdown Detection in Dialogue Breakdown Detection Challenge 3
    Yuiko Tsunomori; Ryuichiro Higashinaka; Tetsuro Takahashi; Michimasa Inaba
    Transactions of the Japanese Society for Artificial Intelligence, Japanese Society for Artificial Intelligence, 35巻, 1号, 掲載ページ DSI-G_1, 出版日 2020年01月01日
    研究論文(学術雑誌)
  • Dialogue system live competition: identifying problems with dialogue systems through live event
    Ryuichiro Higashinaka; Kotaro Funakoshi; Michimasa Inaba; Yuiko Tsunomori; Tetsuro Takahashi; Reina Akama
    Proceedings of the Tenth International Workshop on Spoken Dialogue Systems Technology (IWSDS), Springer, 掲載ページ 185-199, 出版日 2019年04月, 査読付
    研究論文(国際会議プロシーディングス)
  • Overview of the Dialogue Breakdown Detection Challenge 4.
    Ryuichiro Higashinaka; Luis F. D'Haro; Bayan Abu Shawar; Rafael E. Banchs; Kotaro Funakoshi; Michimasa Inaba; Yuiko Tsunomori; Tetsuro Takahashi; João Sedoc
    Increasing Naturalness and Flexibility in Spoken Dialogue Interaction - 10th International Workshop on Spoken Dialogue Systems(IWSDS), Springer, 掲載ページ 403-417, 出版日 2019年
    研究論文(国際会議プロシーディングス)
  • A Method for Identification of Students' States Using Kinect.
    Takashi Ito; Kenta Kamiya; Kenichi Takahashi; Michimasa Inaba
    Proceedings of the 13th International Conference on Ubiquitous Information Management and Communication(IMCOM), Springer, 掲載ページ 341-350, 出版日 2019年
    研究論文(国際会議プロシーディングス)
  • Overview of the sixth dialog system technology challenge: DSTC6.
    Chiori Hori; Julien Perez; Ryuichiro Higashinaka; Takaaki Hori; Y-Lan Boureau; Michimasa Inaba; Yuiko Tsunomori; Tetsuro Takahashi; Koichiro Yoshino; Seokhwan Kim
    Comput. Speech Lang., 55巻, 掲載ページ 1-25, 出版日 2019年, 査読付
    研究論文(学術雑誌)
  • 雑談対話システムにおける対話破綻検出器の評価尺度の検討
    角森 唯子; 東中 竜一郎; 高橋 哲朗; 稲葉 通将
    人工知能学会研究会資料 言語・音声理解と対話処理研究会, 一般社団法人 人工知能学会, 84巻, 掲載ページ 06, 出版日 2018年10月15日
    日本語
  • Improving the Performance of Chat-Oriented Dialogue Systems via Dialogue Breakdown Detection.
    Michimasa Inaba; Kenichi Takahashi
    9th International Workshop on Spoken Dialogue System Technology(IWSDS), Springer, 掲載ページ 345-356, 出版日 2018年
    研究論文(国際会議プロシーディングス)
  • Estimating User Interest from Open-Domain Dialogue.
    Michimasa Inaba; Kenichi Takahashi
    Proceedings of the 19th Annual SIGdial Meeting on Discourse and Dialogue, Melbourne, Australia, July 12-14, 2018, Association for Computational Linguistics, 掲載ページ 32-40, 出版日 2018年, 査読付
    研究論文(国際会議プロシーディングス)
  • Answering What-type and Who-type Questions for Non-task-oriented Dialogue Agents.
    Makoto Koshinda; Michimasa Inaba; Kenichi Takahashi
    Proceedings of the 10th International Conference on Agents and Artificial Intelligence, ICAART 2018, Volume 2, Funchal, Madeira, Portugal, January 16-18, 2018., SciTePress, 掲載ページ 287-294, 出版日 2018年, 査読付
    研究論文(国際会議プロシーディングス)
  • 複数の木を持つGPにおける自動化のための手法
    伊藤崇; 高橋健一; 稲葉通将
    知能と情報(Web), 30巻, 3号, 掲載ページ 571‐580(J‐STAGE), 出版日 2018年, 査読付
    日本語
  • Generating human-like discussion by paraphrasing a translation by the AIWolf protocol using werewolf BBS logs
    Hirofumi Nakamura; Daisuke Katagami; Fujio Toriumi; Hirotaka Osawa; Michimasa Inaba; Kosuke Shinoda; Yoshinobu Kano
    IEEE International Conference on Fuzzy Systems, Institute of Electrical and Electronics Engineers Inc., 掲載ページ 1-6, 出版日 2017年08月23日, 査読付, 'Are you a werewolf?' is one of the most popular communication games and is played globally. The AIWolf Project developed an agent, named 'the AIWolf,' that can play 'Areyou a werewolf?'. An AIWolf utters its thoughts using an AIWolf Protocol. As it is difficult for humans to understand the AIWolf Protocol, translation into natural language is required when human players are involved. However, the conventional method of translation uses a word-to-word method, creating the impression that the utterances have been generated by a machine. This study aimed make the utterances of AIWolf sound more human. The authors set the target that a human player would be unable to distinguish human speech from that generated by AIWolves (the Turing test). The authors define the situation as the maximum value of humanity. The output of translated AIWolf Protocol was paraphrased using data from Werewolf BBS Logs. This study considers making the utterances of AIWolf sound more human using Werewolf BBS Logs and a possibility assignment equation with fuzzy sets. In this paper, an experiment was conducted to confirm whether paraphrasing the utterances of AIWolf using Werewolf BBS Logs for human-like speech is useful or not. It was shown that the experimental method produced slightly more human-like speech than the conventional method.
    研究論文(国際会議プロシーディングス), 英語
  • 非タスク指向型対話システムのためのリカレントニューラルネットワークを用いた発話候補ランキング
    稲葉通将; 高橋健一
    電子情報通信学会論文誌 D(Web), J100-D巻, 6号, 掲載ページ 661‐671 (WEB ONLY), 出版日 2017年06月, 査読付
    日本語
  • AI Wolf Contest - Development of Game AI using Collective Intelligence -
    大澤, 博隆; Toriumi, Fujio; Inaba, Michimasa; Katagami, Daisuke; Shinoda, Kosuke; Matsubara, Hitoshi
    Workshop on Computer Games International Workshop on General Intelligence in Game-Playing Agents CCIS, 掲載ページ 101-115, 出版日 2017年04月, 査読付
    研究論文(学術雑誌), 英語
  • Experiments on images displayed for praising and scolding in e-learning systems
    Kenichi Takahashi; Masakazu Takeue; Michimasa Inaba
    Proceedings of the 11th International Conference on Ubiquitous Information Management and Communication, IMCOM 2017, Association for Computing Machinery, Inc, 掲載ページ 12, 出版日 2017年01月05日, 査読付, Currently, in addition to cloud computing and broadband communication, e-learning is being provided through various devices such as smart phones and tablets. While e-learning has many advantages such as users being able to study everywhere and at any time until they understand the study contents, it also has disadvantages such as users easily becoming bored of just reading computer displays without being able to communicate with teachers or to ask questions. This paper presents the results of learning experiments on maintaining users' will to learn. Functions to display images for praising and scolding users during learning are added to a web-based e-learning system constructed for learning English words. Previous research has showed that an e-leaning system displaying praising and scolding images is effective for maintaining learning will, and images of actors and friends are better than those of a teacher for maintaining learning will. In this study, the system is extended to display favorable images for praising and disgusting images for scolding, and images of anime characters and animals are added to those of actors. Experiments displaying favorable images for scolding as well as praising are conducted to determine if the system can maintain learning will better. After the experiments, a questionnaire is administered to determine the feelings of users themselves.
    研究論文(国際会議プロシーディングス), 英語
  • ゲームプログラミング大会を用いた集合知的ゲームAI開発手法―人狼知能大会の開催と参加エージェントの分析―
    鳥海不二夫; 稲葉通将; 大澤博隆; 片上大輔; 篠田孝祐; 松原仁
    デジタルゲーム学研究, 日本デジタルゲーム学会, 9巻, 1号, 掲載ページ 1‐11-11, 出版日 2016年12月, 査読付
    日本語
  • 雑談ばかりしていると殺される?人狼ゲームにおける発話行為タグセットの提案とプレイヤの行動・勝敗の分析
    稲葉通将; 大畠菜央実; 大畠菜央実; 高橋健一; 鳥海不二夫
    情報処理学会論文誌ジャーナル(Web), 57巻, 11号, 掲載ページ 2392‐2402 (WEB ONLY)-2402, 出版日 2016年11月15日, 本研究では,人狼ゲームにおけるプレイヤの発話内容を表現するタグを設計し,それらのタグを人狼BBSにおけるプレイヤの発話に付与し分析を行った.分析では,襲撃対象,および処刑対象の決定にプレイヤごとの発話の傾向がどのように影響するのか,また,ゲーム全体のコミュニケーションの傾向とゲームの勝敗の関係について調査した.分析の結果,人間側,人狼側の各プレイヤが自陣営の勝利のために効果的なコミュニケーション戦略,および特定のコミュニケーションとプレイヤの行動の関係が明らかとなった.We focus on a communication game "Werewolf". Our final objective is making a werewolf player agent. In this paper, as a first step, we analyze this game using players' utterances. We define a tag set that describes player's dialogue act and annotate utterances in the Werewolf BBS with these tags. By using these annotated utterances, we analyze the relationships between the trend of the types of the utterances and decision-making and victory or defeat. The results of the analysis clarify effective actions in werewolf in order to win.
    日本語
  • 対話システムが話し手役になるためのオープンドメイン独話生成
    稲葉 通将; 吉野 友香; 高橋 健一
    人工知能学会論文誌, The Japanese Society for Artificial Intelligence, 31巻, 1号, 掲載ページ DSF-F_1-9, 出版日 2016年01月08日, 査読付, This paper presents an open domain monologue generation method for non-task-oriented dialogue systems to be able to speak their opinions and impressions as a speaker. To generate monologues, we acquire suitable utterances that contain a given topic from Twitter. Our method determines whether utterances have cohesion or not using the support vector machine and concatenate them in a row. It scores the utterance sequences from the aspect of their humor, unexpectedness and speciality in the given topic. We acquire the utterance sequences that ranks high as monologues. Results of an experiment demonstrate that our method can generate amusing and semantically appropriate monologues.
    研究論文(学術雑誌), 日本語
  • 対話システムが話し手役になるためのオープンドメイン独話生成
    稲葉 通将; 吉野 友香; 高橋 健一
    人工知能学会論文誌, The Japanese Society for Artificial Intelligence, 31巻, 1号, 出版日 2016年01月08日, 査読付, This paper presents an open domain monologue generation method for non-task-oriented dialogue systems to be able to speak their opinions and impressions as a speaker. To generate monologues, we acquire suitable utterances that contain a given topic from Twitter. Our method determines whether utterances have cohesion or not using the support vector machine and concatenate them in a row. It scores the utterance sequences from the aspect of their humor, unexpectedness and speciality in the given topic. We acquire the utterance sequences that ranks high as monologues. Results of an experiment demonstrate that our method can generate amusing and semantically appropriate monologues.
    研究論文(学術雑誌), 日本語
  • 人狼知能大会における統計分析とSVMを用いた人狼推定を行うエージェントの設計
    梶原, 健吾; 鳥海, 不二夫; 稲葉, 通将; 大澤, 博隆; 片上, 大輔; 篠田, 孝祐; 松原, 仁; 狩野, 芳伸
    JSAI大会論文集, 一般社団法人 人工知能学会, 2016巻, 掲載ページ 2F41-2F41, 出版日 2016年
    研究論文(研究会,シンポジウム資料等), 日本語
  • Constructing a Human-like agent for the Werewolf Game using a psychological model based multiple perspectives
    Noritsugu Nakamura; Michimasa Inaba; Kenichi Takahashi; Fujio Toriumi; Hirotaka Osawa; Daisuke Katagami; Kousuke Shinoda
    PROCEEDINGS OF 2016 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (SSCI), IEEE, 掲載ページ 1-8, 出版日 2016年, 査読付, In this paper, we focus on the Werewolf Game. The Werewolf Game is an advanced communication-game in which winning or losing is directly linked to one's success or failure in communication. Therefore, we expect exponential developments in artificial intelligence by studying the Werewolf Game. In this current study, we propose a psychological model that considers multiple perspectives to model the play of a human such as inferring the intention of the other side. As one of the psychological models, we constructed a "one's self model" that models the role of others as viewed from their own viewpoint. In addition, to determine whether one's opinion is reliable after inferring other's intentions, we also constructed an "others model" that models the role of others as viewed from their viewpoints. Combining these models, we showed through experimentation that a combined approach achieved better results, i.e., higher win percentages.
    研究論文(国際会議プロシーディングス), 英語
  • Backchanneling via Twitter Data for Conversational Dialogue Systems
    Michimasa Inaba; Kenichi Takahashi
    Speech and Computer, SPRINGER INT PUBLISHING AG, 9811巻, 掲載ページ 148-155, 出版日 2016年, 査読付, Backchanneling plays a crucial role in human-to-human communication. In this study, we propose a method for generating a rich variety of backchanneling, which is not just limited to simple "hm" or "sure" responses, to realize smooth communication in conversational dialogue systems. We formulate the problem of what the backchanneling generation function should return for given user inputs as a multi-class classification problem and determine a suitable class using a recurrent neural network with a long short-term memory. Training data for our model comprised pairs of tweets and replies acquired from Twitter. Experimental results demonstrated that our method can appropriately select backchannels to given inputs and significantly outperform baseline methods.
    研究論文(国際会議プロシーディングス), 英語
  • Neural Utterance Ranking Model for Conversational Dialogue Systems.
    Michimasa Inaba; Kenichi Takahashi
    Proceedings of the SIGDIAL 2016 Conference, The 17th Annual Meeting of the Special Interest Group on Discourse and Dialogue, 13-15 September 2016, Los Angeles, CA, USA, The Association for Computer Linguistics, 掲載ページ 393-403, 出版日 2016年, 査読付
    研究論文(国際会議プロシーディングス)
  • The dialogue breakdown detection challenge: Task description, datasets, and evaluation metrics.
    Ryuichiro Higashinaka; Kotaro Funakoshi; Yuka Kobayashi; Michimasa Inaba
    Proceedings of the Tenth International Conference on Language Resources and Evaluation LREC 2016, Portorož, Slovenia, May 23-28, 2016., European Language Resources Association (ELRA), 出版日 2016年, 査読付
    研究論文(国際会議プロシーディングス)
  • Construction of a Player Agent for a Card Game Using an Ensemble Method
    Kenta Nimoto; Kenichi Takahashi; Michimasa Inaba
    KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS: PROCEEDINGS OF THE 20TH INTERNATIONAL CONFERENCE KES-2016, ELSEVIER SCIENCE BV, 96巻, 掲載ページ 772-781, 出版日 2016年, 査読付, The 3-channel fuzzy ART network FALCON (Fusion Architecture for Learning, COgnition, and Navigation) is known as an effective method for combining reinforcement learning with state segmentation. It has been shown that FALCON is effective in making a player agent for the card game Hearts, although the agent was unable to beat an agent using the UCT algorithm developed for Monte-Carlo simulation. This study proposes an ensemble method for FALCON to make an agent stronger. The method uses nine types of learners and combines them to decide an action. Experiments demonstrate that our approach is superior to an agent using a single learner. (C) 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of KES International
    研究論文(国際会議プロシーディングス), 英語
  • Interactive learning of a FALCON for a card game
    Kazuma Kasahara; Kenta Nimoto; Kenichi Takahashi; Michimasa Inaba
    KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS: PROCEEDINGS OF THE 20TH INTERNATIONAL CONFERENCE KES-2016, ELSEVIER SCIENCE BV, 96巻, 掲載ページ 129-138, 出版日 2016年, 査読付, Among many reinforcement learning methods, FALCON is a machine learning method which is an extend fuzzy ART( Adaptive Resonance Theory), and can appropriately discretize a state space. FALCON is an on-line method proposed by Ah-Hwee Tan. It can discretize a state space and learn action rules simultaneously by learning relations among percepts, actions, and rewards. In this study, a learning agent using FALCON is interactively trained, and the learning effect is measured through experiments. In experiments, the learning agent learns by playing 50,000 card games of "Hearts" against three rule-based agents. Then, the interface that agents can interactively play the game with human cooperators is made so that human cooperators can play the game against the learning agent to strengthen it. It continues learning during games. The effectiveness of interactive learning is ascertained through the experiments. (C) 2016 The Authors. Published by Elsevier B.V.
    研究論文(国際会議プロシーディングス), 英語
  • Obtaining repetitive actions for genetic programming with multiple trees
    Takashi Ito; Kenichi Takahashi; Michimasa Inaba
    KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS: PROCEEDINGS OF THE 20TH INTERNATIONAL CONFERENCE KES-2016, ELSEVIER SCIENCE BV, 96巻, 掲載ページ 120-128, 出版日 2016年, 査読付, This paper proposes a method to improve genetic programming with multiple trees (GPCN). An individual in GPCN comprises multiple trees, and each tree has a number P that indicates the number of repetitive actions based on the tree. In previous work, a method for updating the number P has been proposed to obtain P suitable to the tree in evolution. However, in the method efficiency becomes worse as the range of P becomes wider. In order to solve the problem, in this study, two methods are proposed: inheriting the number P of a tree from an excellent individual and using mutation for preventing the number P from being into a local optimum. Additionally, a method to eliminate trees consisting of a single terminal node is proposed. (C) 2016 The Authors. Published by Elsevier B.V.
    研究論文(国際会議プロシーディングス), 英語
  • Experimental Investigation for a Human Relationship Formation Support Agent using Information Presentation During Conversation.
    Michimasa Inaba; Kana Otsuka; Kenichi Takahashi
    Proceedings of the 8th International Conference on Agents and Artificial Intelligence (ICAART 2016), Volume 1, Rome, Italy, February 24-26, 2016., SciTePress, 掲載ページ 87-94, 出版日 2016年, 査読付
    研究論文(国際会議プロシーディングス)
  • Werewolf game modeling using action probabilities based on play log analysis
    Yuya Hirata; Michimasa Inaba; Kenichi Takahashi; Fujio Toriumi; Hirotaka Osawa; Daisuke Katagami; Kousuke Shinoda
    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Springer Verlag, 10068巻, 掲載ページ 103-114, 出版日 2016年, 査読付, In this study, we construct a non-human agent that can play the werewolf game (i.e., AI wolf) with aims of creating more advanced intelligence and acquire more advanced communication skills for AIbased systems. We therefore constructed a behavioral model using information regarding human players and the decisions made by such players
    all such information was obtained from play logs of the werewolf game. To confirm our model, we conducted simulation experiments of the werewolf game using an agent based on our proposed behavioral model, as well as a random agent for comparison. Consequently, we obtained an 81.55% coincidence ratio of agent behavior versus human behavior.
    研究論文(国際会議プロシーディングス), 英語
  • Improvement of Agent Learning for a Card Game Based on Multi-channel ART Networks.
    Kenta Nimoto; Kenichi Takahashi; Michimasa Inaba
    JCP, 11巻, 4号, 掲載ページ 341-352, 出版日 2016年, 査読付
  • Extension of Genetic Programming with Multiple Trees for Agent Learning.
    Takashi Ito; Kenichi Takahashi; Michimasa Inaba
    JCP, 11巻, 4号, 掲載ページ 329-340, 出版日 2016年, 査読付
  • 大学生アンケートからの文系理系学生の特徴に関する分析
    野津田 雄太; 高橋 健一; 稲葉 通将
    情報処理学会論文誌教育とコンピュータ(TCE), 1巻, 4号, 掲載ページ 83-92, 出版日 2015年12月, 査読付, 1990年頃から国内で理系離れが問題視されている.しかし,理系離れの要因として明確なものは存在しない.その分析に寄与するため,文系・理系の大学生の日常生活の習慣や科目の得意・不得意などの状況を調査した.本研究では,著者が所属する大学の学生を対象にアンケート調査を実施し,データマイニング手法を用いて,科目の履修・嗜好などの傾向から文系・理系への進路選択と相関のある要因と,学生の生活習慣から文系・理系学生の日常の傾向について考察を行う.用いたデータマイニング手法は決定木,主成分分析,ベイジアンネットワークである.分析の結果,小学校・中学校からの得意・不得意科目や日常生活における学生の嗜好と学生の文理に相関があることが分かった.From around the 1990s, undergraduate students majoring in sciences have decreased, and indifference to sciences has been spread. However, specific factors of the indifference to sciences are not clear. In order to contribute to the analysis, investigation on lifestyle and subjects studied in schools by students majoring in humanities and sciences is done. This paper describes experimental results to extract features of undergraduates majoring in humanities and sciences respectively to show factors that correlate with their course, i.e., humanities or sciences. In the research, data are collected through questionnaires to undergraduates. The questionnaire includes questions such as the lifestyle, subjects studied in schools. Then the data are analyzed by using data mining methods, namely decision trees, Principal Component Analysis, and Bayesian network. The analysis shows a correlation between students in sciences and humanities and lifestyle and a correlation between those students and subjects studied in elementary and junior high schools.
    日本語
  • Movement design of a life-like agent for the werewolf game
    Daisuke Katagami; Masashi Kanazawa; Fujio Toriumi; Hirotaka Osawa; Michimasa Inaba; Kosuke Shinoda
    IEEE International Conference on Fuzzy Systems, Institute of Electrical and Electronics Engineers Inc., 2015-巻, 出版日 2015年11月25日, 査読付, In this research, we target at the interactive communication game 'werewolf' with a subject of research. Werewolf is a popular party game all over the world, and the relevance studies have been advanced in recent years. However, the life-like agent who does werewolf has not been developed. Therefore the purpose of this research is to analyze non-verbal information from movies which play the werewolf with face-to-face communication and to make clear the impression for others by the movements of players in the game. Moreover, we verify whether the life-like agent gives an impression like human in werewolf game by mounting the movements on a life-like agent.
    研究論文(国際会議プロシーディングス), 英語
  • 人狼知能プロジェクト(<特集>エンターテイメントにおけるAI)
    片上, 大輔; 鳥海, 不二夫; 大澤, 博隆; 稲葉, 通将; 篠田, 孝祐; 松原, 仁
    人工知能:人工知能学会誌, 社団法人人工知能学会, 30巻, 1号, 掲載ページ 65-73, 出版日 2015年01月
    研究論文(学術雑誌), 日本語
  • Statistical Response Method and Learning Data Acquisition using Gamified Crowdsourcing for a Non-task-oriented Dialogue Agent
    Michimasa Inaba; Naoyuki Iwata; Fujio Toriumi; Takatsugu Hirayama; Yu Enokibori; Kenichi Takahashi; Kenji Mase
    AGENTS AND ARTIFICIAL INTELLIGENCE, ICAART 2014, SPRINGER-VERLAG BERLIN, 8946巻, 掲載ページ 119-136, 出版日 2015年, 査読付, This paper presents a proposal of a construction method for non-task-oriented dialogue agents (chatbots) that are based on the statistical response method. The method prepares candidate utterances in advance. From the data, it learns which utterances are suitable for context. Therefore, a dialogue agent constructed using our method automatically selects a suitable utterance depending on a context from candidate utterances. This paper also proposes a low-cost quality-assured method of learning data acquisition for the proposed response method. The method uses crowdsourcing and brings game mechanics to data acquisition.
    Results of an experiment using learning data obtained using the proposed data acquisition method demonstrate that the appropriate utterance is selected with high accuracy.
    研究論文(国際会議プロシーディングス), 英語
  • Machine-Learned Ranking based Non-task-oriented Dialogue Agent using Twitter Data
    Makoto Koshinda; Michimasa Inaba; Kenichi Takahashi
    2015 IEEE/WIC/ACM INTERNATIONAL CONFERENCE ON WEB INTELLIGENCE AND INTELLIGENT AGENT TECHNOLOGY (WI-IAT), VOL 3, IEEE COMPUTER SOC, 掲載ページ 5-8, 出版日 2015年, 査読付, This paper describes a method for developing a non-task-oriented dialogue agent (also called chat-oriented or conversational dialogue agents) that can cover broad range of topics. Our method extracts a topic from a user's utterance and acquires candidate utterances that contain the topic from Twitter. Our agent selects a suitable utterance for dialogue context from candidates using machine-learned ranking method. Results of an experiment demonstrate that a dialogue agent based on the proposed method can conduct more natural and enjoyable conversation compared to other dialogue agents.
    研究論文(国際会議プロシーディングス), 英語
  • Movement design of a life-like agent for the werewolf game
    Daisuke Katagami; Masashi Kanazawa; Fujio Toriumi; Hirotaka Osawa; Michimasa Inaba; Kosuke Shinoda
    2015 IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS (FUZZ-IEEE 2015), IEEE, 掲載ページ 1-7, 出版日 2015年, 査読付, In this research, we target at the interactive communication game "werewolf' with a subject of research. Werewolf is a popular party game all over the world, and the relevance studies have been advanced in recent years. However, the life-like agent who does werewolf has not been developed. Therefore the purpose of this research is to analyze non-verbal information from movies which play the werewolf with face-to-face communication and to make clear the impression for others by the movements of players in the game. Moreover, we verify whether the life-like agent gives an impression like human in werewolf game by mounting the movements on a life-like agent.
    研究論文(国際会議プロシーディングス), 英語
  • Development of werewolf match system for human players mediated with lifelike agents
    Yu Kobayashi; Hirotaka Osawa; Michimasa Inaba; Kosuke Shinoda; Fujio Toriumi; Daisuke Katagami
    HAI 2014 - Proceedings of the 2nd International Conference on Human-Agent Interaction, Association for Computing Machinery, Inc, 掲載ページ 205-207, 出版日 2014年10月29日, 査読付, "Are You a Werewolf?" is a conversation type game. We construct a Werewolf match system for humans with lifelike agents. We evaluate whether it is possible to realize conversation space of "Are You a Werewolf?" with the system.
    研究論文(国際会議プロシーディングス), 英語
  • Investigation of the effects of nonverbal information on werewolf
    Daisuke Katagami; Shono Takaku; Michimasa Inaba; Hirotaka Osawa; Kosuke Shinoda; Junji Nishino; Fujio Toriumi
    IEEE International Conference on Fuzzy Systems, Institute of Electrical and Electronics Engineers Inc., 掲載ページ 982-987, 出版日 2014年09月04日, 査読付, Werewolf is one of the popular communication games all over the world. It treats ambiguity of human discussion including the utterances, gestures and facial expressions in a broad sense. In this research, we pay attention to this imperfect information game werewolf. The purpose of the research is to develop an intelligent agent 'AI werewolf' which is enabled to naturally play werewolf with human. This paper aims to investigate how behavior contribute to victory of own-side players by using machine learning as a first step. As the results of investigation and analysis of the playing movie, we found that nonverbal information in the game of werewolf has importance to winning or losing the game.
    研究論文(国際会議プロシーディングス), 英語
  • 人狼人工知能プロジェクトの概要と展望
    鳥海, 不二夫; 稲葉, 通将; 大澤, 博隆; 片上, 大輔; 篠田, 孝祐; 松原, 仁
    情報処理学会研究報告. GI, [ゲーム情報学], 一般社団法人情報処理学会, 2014巻, 5号, 掲載ページ 1, 出版日 2014年06月, 不完全情報コミュニケーションゲームである「人狼」を題材にした研究プロジェクトの概要を紹介し,ゲーム情報学という観点からこのプロジェクトの将来について展望していく.
    研究論文(学術雑誌), 日本語
  • Twitterを用いた非タスク指向型対話システムのための発話候補文獲得
    稲葉 通将; 神園 彩香; 高橋 健一
    人工知能学会論文誌, The Japanese Society for Artificial Intelligence, 29巻, 1号, 掲載ページ 21-31, 出版日 2014年, 査読付, Recently, computerized dialogue systems are studied actively. Non-task-oriented dialogue systems that handle domain-free dialogues like chats are expected be applied in various fields, but many challenges still exist in developing them. This paper addresses the problem of utterance generation for non-task-oriented dialogue systems. We search twitter data by topic words and acquire sentences. The sentences are filtered by rules and scored on the basis of training data. We acquire the sentences which have a high score as utterances. The results of an experiment demonstrate that the proposed method can generate appropriate utterances with a high degree of accuracy.
    研究論文(学術雑誌), 日本語
  • Experiments Assessing Learning of Agent Behavior using Genetic Programming with Multiple Trees.
    Takashi Ito; Kenichi Takahashi; Michimasa Inaba
    ICAART 2014 - Proceedings of the 6th International Conference on Agents and Artificial Intelligence, Volume 1, ESEO, Angers, Loire Valley, France, 6-8 March, 2014, SciTePress, 掲載ページ 264-271, 出版日 2014年, 査読付
  • Constructing a Non-task-oriented Dialogue Agent using Statistical Response Method and Gamification.
    Michimasa Inaba; Naoyuki Iwata; Fujio Toriumi; Takatsugu Hirayama; Yu Enokibori; Kenichi Takahashi; Kenji Mase
    ICAART 2014 - Proceedings of the 6th International Conference on Agents and Artificial Intelligence, Volume 1, ESEO, Angers, Loire Valley, France, 6-8 March, 2014, SciTePress, 掲載ページ 14-21, 出版日 2014年, 査読付
  • Investigation of the Effects of Nonverbal Information on Werewolf
    Daisuke Katagami; Shono Takaku; Michimasa Inaba; Hirotaka Osawa; Kosuke Shinoda; Junji Nishino; Fujio Toriumi
    2014 IEEE INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS (FUZZ-IEEE), IEEE, 掲載ページ 982-987, 出版日 2014年, 査読付, Werewolf is one of the popular communication games all over the world. It treats ambiguity of human discussion including the utterances, gestures and facial expressions in a broad sense. In this research, we pay attention to this imperfect information game werewolf. The purpose of the research is to develop an intelligent agent "AI werewolf" which is enabled to naturally play werewolf with human. This paper aims to investigate how behavior contribute to victory of own-side players by using machine learning as a first step. As the results of investigation and analysis of the playing movie, we found that nonverbal information in the game of werewolf has importance to winning or losing the game.
    研究論文(国際会議プロシーディングス), 英語
  • Strategy Selection by Reinforcement Learning for Multi-Car Elevator Systems
    Masaki Ikuta; Kenichi Takahashi; Michimasa Inaba
    2013 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS (SMC 2013), IEEE, 掲載ページ 2479-2484, 出版日 2013年, 査読付, This paper discusses the group control of elevators for improving efficiency; an efficient control method for multi-car elevator using reinforcement learning is proposed. In the method, the control agent selects the best strategy among four strategies, namely Transportation strategy, Passenger strategy, Zone strategy, and Difference strategy according to traffic flow. The control agent takes the number of total passengers and the distance from the departure floor to the destination floor of a call into account. Through experiments, the performance of the proposed method is shown; the average service time of the proposed method is compared with the average service time obtained for the cases where the car assignment is made by each of the three or four strategies.
    研究論文(国際会議プロシーディングス), 英語
  • Effects of images displayed for praising or scolding on motivation in E-learning
    Masakazu Takeue; Kenichi Takahashi; Michimasa Inaba
    Proceedings - 2013 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2013, IEEE, 掲載ページ 2073-2078, 出版日 2013年, 査読付, E-learning systems that use computers and the internet have become popular and ubiquitous. E-learning systems offer many benefits. However, users often lose their motivation for learning during the process of studying, and the frequency with which they use e-learning systems sometimes decreases. Results show that a function to praise or scold users by displaying images of a teacher is effective to improve or maintain their motivation. In this study, that function is modified to display images that users like or dislike, respectively, for praising and scolding learners. Through experimentation, this study assesses the effects of displayed images on maintaining the will to learn. © 2013 IEEE.
    研究論文(国際会議プロシーディングス), 英語
  • 非タスク指向型対話エージェントのための統計的応答手法
    稲葉 通将; 平井 尚樹; 鳥海 不二夫; 石井 健一郎
    電子情報通信学会論文誌. D, 情報・システム, 一般社団法人電子情報通信学会, 95巻, 6号, 掲載ページ 1390-1400, 出版日 2012年06月, 査読付, 近年,対話エージェント研究は特定のタスク達成を目的としたタスク指向型対話エージェントから,達成すべきタスクをもたず,雑談などを扱う非タスク指向型対話エージェントへと広がりつつある.本論文では,非タスク指向型対話エージェントの応答手法の新たな試みとして,ランキング学習を用いた統計的応答手法を提案する.ランキング学習とは,情報検索の分野で研究が進められている技術であり,複数のデータの中から目的のデータを求める際に,目的のデータが上位に順位付けされるよう学習を行う統計的機械学習法である.本研究では,あらかじめ大量の発話候補を用意しておき,その発話候補を,ランキング学習を用いてそれまでの対話の流れに対する適切さによって順位付けし,最も上位に順位付けされた発話をエージェントの応答として選択する手法を提案する.
    日本語
  • Experiments of Displaying Images to Keep the Motivation in E-learning
    Masakazu Takeue; Kazutoshi Shimada; Kenichi Takahashi; Michimasa Inaba
    PROCEEDINGS 2012 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN, AND CYBERNETICS (SMC), IEEE, 掲載ページ 120-125, 出版日 2012年, 査読付, E-learning systems that use computers and the Internet have become popular. E-learning systems have many advantages. However, the users often lose their motivation for learning in the process of studying, and the frequency that they use e-learning systems sometimes decreases. In order to improve or keep their motivation, we add a function that the users are praised or scolded by displaying images of a teacher, actors, and friends. We check to see if the function is effective or not through experiments.
    研究論文(国際会議プロシーディングス), 英語
  • 語の共起情報を用いた対話における盛り上がりの自動判定(ヒューマンコミュニケーション基礎,<特集>ヒューマンコミュニケーション〜人間中心の情報環境構築のための要素技術〜論文)
    稲葉 通将; 鳥海 不二夫; 石井 健一郎
    電子情報通信学会論文誌. D, 情報・システム, 一般社団法人電子情報通信学会, 94巻, 1号, 掲載ページ 59-67, 出版日 2011年01月, 査読付, 近年,人間と対話を行うコンピュータ(対話エージェント)に関する研究が活発化している.しかし,その多くはチケット予約や道案内など特定のタスク達成を目的としたタスク指向型対話エージェントに関するものであり,人間と雑談を行い,話を盛り上げるための非タスク指向型対話エージェントの研究は少ない.話を盛り上げ,相手を楽しませる対話エージェントの設計のためには,人間同士の対話の分析が必要不可欠である.そこで本論文では,人間同士のテキスト対話中における,話者の盛り上がり度の高い発話をCRF(Conditional Random Fields)を用いて自動判定する手法を提案する.本手法では,大規模コーパスから獲得した共起情報を用いて,発話と発話の間の意味的なつながりの強さを抽出し,これを素性の一つとして用いた.実験の結果,提案手法は人手による評価に準ずる性能を示すことを確認した.
    日本語
  • 語の共起情報を用いた非タスク指向型対話エージェントの評価(ヒューマンコミュニケーション基礎,<特集>ヒューマンコミュニケーション〜人間中心の情報環境構築のための要素技術〜論文)
    稲葉 通将; 鳥海 不二夫; 石井 健一郎
    電子情報通信学会論文誌. D, 情報・システム, 一般社団法人電子情報通信学会, 94巻, 1号, 掲載ページ 48-58, 出版日 2011年01月, 査読付, 近年,コンピュータによるコミュニケーション支援の研究が活発化しており,その一つが対話エージェントに関する研究である.しかし,その多くはチケット予約や道案内など特定のタスク達成を目的としたタスク指向型対話エージェントに関するものであり,人間と雑談をし,話を盛り上げるための非タスク指向型対話エージェントの研究は少ない.このようなエージェントを実現するためには,客観的・定量的な評価法が不可欠である.本論文では,ある発話が周囲の文脈との間に語彙的結束性を有する場合,その発話は意味的に正しい発話であるという仮定に基づき,対話の意味に踏み込んだ評価手法を提案する.実験の結果,複数の非タスク指向型対話エージェントの性能を比較評価できることが確認された.また,人間による評価との相関を有することが確認された.
    日本語
  • Automatic detection of enthusiasm in non-task-oriented dialogues using word co-occurrence
    Michimasa Inaba; Fujio Toriumi; Kenichiro Ishii
    IEEE SSCI 2011 - Symposium Series on Computational Intelligence - WACI 2011: 2011 Workshop on Affective Computational Intelligence, IEEE, 掲載ページ 62-67, 出版日 2011年, 査読付, A method is proposed for automatically detecting "enthusiastic" utterances in text-based dialogues. Using conditional random fields, our proposed method distinguishes between the enthusiastic and non-enthusiastic parts of a dialogue. Testing demonstrated that it performs as well as human detection. Being able to distinguish between the enthusiastic and non-enthusiastic parts makes it possible to quantitatively analyze the phenomenon of enthusiasm, which should lead to a practical approach to the creation of non-task-oriented agents that can help generate enthusiastic dialogues. © 2011 IEEE.
    研究論文(国際会議プロシーディングス), 英語

MISC

  • 対話システムライブコンペティション5
    東中 竜一郎; 高橋 哲朗; 堀内 颯太; 稲葉 通将; 佐藤 志貴; 船越 孝太郎; 小室 允人; 西川 寛之; 宇佐美 まゆみ; 港 隆史; 境 くりま; 船山 智
    対話システムライブコンペティション5の概要について述べる.具体的には,マルチモーダル対話を対象にした背景,オープントラックとシチュエーショントラックのタスク設定,評価尺度などについて説明し,予選の結果について述べる., 一般社団法人 人工知能学会, 出版日 2022年12月01日, 人工知能学会研究会資料 言語・音声理解と対話処理研究会, 96巻, 掲載ページ 19, 日本語, 0918-5682, 2436-4576, AN10432166
  • 多様な年代の話者による旅行代理店タスク対話コーパスの収集と分析
    稲葉 通将; 東中 竜一郎; 千葉 祐弥; 駒谷 和範; 宮尾 祐介; 長井 隆行
    人が他者とコミュニケーションを行う際,語彙や話す速さ,敬語の有無などを相手に応じて使い分けている.しかし,現在の対話システムでは,ユーザに応じて話し方を変更することはほとんどなされていない.対話システムがより効率的にタスクを達成したり,ユーザの満足度をより高めるためには,ユーザに応じて対話戦略などを変更することが望ましい.我々は,話し方の変化に大きく影響を与える要素として話者の年齢に着目し,児童から高齢者まで幅広い年齢層の話者によるマルチモーダル対話コーパスを収集した.対話内容として,幅広い年代が興味を持つ旅行に着目し,旅行代理店における観光相談を元にしたタスクを設定した.本稿では,タスクの詳細,対話データの収集方法と収集結果,および収集したデータの分析結果について述べる., 一般社団法人 人工知能学会, 出版日 2021年11月20日, 人工知能学会研究会資料 言語・音声理解と対話処理研究会, 93巻, 掲載ページ 192-197, 日本語, 0918-5682, 2436-4576, conf/lrec/InabaCHKMN22, 130008119213, AN10432166
  • 対話システムライブコンペティション4
    東中 竜一郎; 船越 孝太郎; 高橋 哲朗; 稲葉 通将; 赤間 怜奈; 佐藤 志貴; 堀内 颯太; ドルサ テヨルス; 小室 允人; 西川 寛之; 宇佐美 まゆみ
    本稿では,対話システムライブコンペティション4について述べる.我々は対話システムを聴衆の前で動作させ,鑑賞・評価を行うイベントである対話ライブコンペティションを2018年から実施してきた.今回で本イベントは4回目となる.任意の話題で雑談を行う能力を評価する「オープントラック」と所定のシチュエーションで人間らしい対話を行う能力を評価する「シチュエーショントラック」がある.本稿では,本イベントの仕様を述べるとともに,各トラックにエントリされたシステムの概要や予選結果について述べる.本選は,第12回対話システムシンポジウムのセッションとして開催予定である., 一般社団法人 人工知能学会, 出版日 2021年, 人工知能学会研究会資料 言語・音声理解と対話処理研究会, 93巻, 掲載ページ 92-100, 日本語, 0918-5682, 2436-4576, 130008119314, AN10432166
  • 対話システムライブコンペティション(3) (第11回対話システムシンポジウム)
    東中 竜一郎; 船越 孝太郎; 高橋 哲朗; 稲葉 通将; 角森 唯子; 赤間 怜奈; 宇佐美 まゆみ; 川端 良子; 水上 雅博; 小室 允人; テヨルス ドルサ
    人工知能学会, 出版日 2020年11月30日, 言語・音声理解と対話処理研究会, 90巻, 掲載ページ 96-103, 日本語, 0918-5682, 40022425648, AN10432166
  • 人狼知能大会 ─国際大会と自然言語大会を終えて─
    鳥海 不二夫; 狩野 芳伸; 大槻 恭士; 大澤 博隆; アランニャ クラウス; 稲葉 通将; 片上 大輔
    一般社団法人 人工知能学会, 出版日 2020年, 人工知能, 35巻, 3号, 掲載ページ 344-350, 日本語, 2188-2266, 130007917804, AA12652467
  • 対話システムライブコンペティションから何が得られたか
    東中 竜一郎; 船越 孝太郎; 稲葉 通将; 角森 唯子; 高橋 哲朗; 赤間 怜奈; 宇佐美 まゆみ; 川端 良子; 水上 雅博
    一般社団法人 人工知能学会, 出版日 2020年, 人工知能, 35巻, 3号, 掲載ページ 333-343, 日本語, 2188-2266, 130007917805, AA12652467
  • 対話システムライブコンペティション(2) (第10回対話システムシンポジウム)
    東中 竜一郎; 船越 孝太郎; 稲葉 通将; 角森 唯子; 高橋 哲朗; 赤間 怜奈; 宇佐美 まゆみ; 川端 良子; 水上 雅博
    人工知能学会, 出版日 2019年12月02日, 言語・音声理解と対話処理研究会, 87巻, 掲載ページ 42-49, 日本語, 0918-5682, 40022078658, AN10432166
  • 人狼知能と不完全ゲーム. 特集:「2019 年度人工知能学会全国大会(第33 回)」
    稲葉 通将; 片上 大輔; 狩野 芳伸; 大槻 恭士
    出版日 2019年11月01日, 人工知能学会誌, 34巻, 6号, 掲載ページ 876-880
  • Introduction
    Yoshinobu Kano; Claus Aranha; Michimasa Inaba; Fujio Toriumi; Hirotaka Osawa; Daisuke Katagami; Takashi Otsuki
    出版日 2019年, AIWolfDial 2019 - Proceedings of the 1st International Workshop of AI Werewolf and Dialog System, Proceedings of the Workshop, 掲載ページ III, 85092009358
  • 人狼知能研究のすすめ
    片上大輔; 鳥海不二夫; 大澤博隆; 狩野芳伸; 稲葉通将; 大槻恭士
    出版日 2018年10月01日, 知能と情報, 2019巻, 05号, 掲載ページ 236-244
  • 人狼知能大会第一回自然言語部門の開催
    狩野芳伸; 稲葉通将
    出版日 2018年, 人工知能学会全国大会論文集(CD-ROM), 32nd巻, 掲載ページ ROMBUNNO.1H2.OS.13b.03, 日本語, 1347-9881, 201802264964629106
  • 人狼BBSに対する役職表明・能力行使報告情報のアノテーション
    稲葉通将; 狩野芳伸; 大澤博隆; 大槻恭士; 片上大輔; 鳥海不二夫
    出版日 2018年, 人工知能学会全国大会論文集(CD-ROM), 32nd巻, 掲載ページ ROMBUNNO.1H1.OS.13a.01, 日本語, 1347-9881, 201802283683820991
  • Overview of Dialogue Breakdown Detection Challenge 3
    Ryuichiro Higashinaka; Kotaro Funakoshi; Michimasa Inaba; Yuiko Tsunomori; Tetsuro Takahashi; Nobuhiro Kaji
    出版日 2017年12月, Dialog System Technology Challenges 6
  • 対話破綻検出による対話システムの応答性能の向上 (第8回対話システムシンポジウム)
    稲葉 通将; 高橋 健一
    人工知能学会, 出版日 2017年10月12日, 言語・音声理解と対話処理研究会, 81巻, 掲載ページ 110-115, 日本語, 0918-5682, 40021358159, AN10432166
  • 人間同士の議論の疑似体験を実現する人狼知能対戦システムの開発
    中村 洋文; 片上 大輔; 鳥海 不二夫; 大澤 博隆; 稲葉 通将; 篠田 孝祐; 狩野 芳伸
    日本知能情報ファジィ学会, 出版日 2017年09月13日, ファジィシステムシンポジウム講演論文集, 33巻, 掲載ページ 383-386, 日本語, 1882-0212, 201702230740673957, 40021612387
  • 複数木構造を持つGPのための木の数の自動決定
    伊藤崇; 高橋健一; 稲葉通将
    出版日 2017年03月16日, 情報処理学会全国大会講演論文集, 79th巻, 2号, 掲載ページ 2.55‐2.56, 日本語, 201702268191413827
  • 人狼知能大会における自然言語処理部門の構想
    狩野芳伸; 稲葉通将
    出版日 2017年, 言語処理学会年次大会発表論文集(Web), 23rd巻, 掲載ページ ROMBUNNO.B1‐5 (WEB ONLY), 日本語, 201702234992009507
  • 雑談対話ログを用いた話者の潜在的興味対象の推定
    縄手優矢; 稲葉通将; 高谷智哉; 山田整; 高橋健一
    出版日 2017年, 人工知能学会全国大会論文集(CD-ROM), 31st巻, 掲載ページ ROMBUNNO.1B2‐OS‐25b‐3, 日本語, 1347-9881, 201802248801971105
  • RNNエンコーダによる文脈を考慮した対話破綻検出 (第7回対話システムシンポジウム)
    稲葉 通将; 高橋 健一
    人工知能学会, 出版日 2016年10月05日, 言語・音声理解と対話処理研究会, 78巻, 掲載ページ 98-101, 日本語, 0918-5682, 40020968137, AN10432166
  • 対話破綻検出チャレンジ(2) (第7回対話システムシンポジウム)
    東中 竜一郎; 船越 孝太郎; 稲葉 通将; 荒瀬 由紀; 角森 唯子
    人工知能学会, 出版日 2016年10月05日, 言語・音声理解と対話処理研究会, 78巻, 掲載ページ 64-69, 日本語, 0918-5682, 40020968045, AN10432166
  • 同乗者エージェントによるドライバー特性とエージェントの印象について (人工知能と知識処理)
    片上 大輔; 佐藤 翔吾; 稲葉 通将; 田中 貴紘; 荒井 良徳
    電子情報通信学会, 出版日 2016年06月27日, 電子情報通信学会技術研究報告 = IEICE technical report : 信学技報, 116巻, 117号, 掲載ページ 1-5, 日本語, 0913-5685, 40020881747, AN10013061
  • Long Short-Term Memory Recurrent Neural Networkを用いた対話破綻検出
    稲葉 通将; 高橋 健一
    人工知能学会, 出版日 2015年10月29日, 言語・音声理解と対話処理研究会, 75巻, 掲載ページ 57-60, 日本語, 0918-5682, 40020633026, AN10432166
  • 人狼知能プロジェクト2015
    篠田 孝祐; 片上 大輔; 稲葉 通将
    人工知能学会, 出版日 2015年10月29日, 言語・音声理解と対話処理研究会, 75巻, 掲載ページ 80-85, 日本語, 0918-5682, 40020633111, AN10432166
  • コミュニケーションゲーム「人狼」を達成するロボットエージェントの設計要件
    大澤博隆; 栢野航; 片上大輔; 鳥海不二夫; 稲葉通将; 篠田孝祐
    出版日 2015年06月22日, 電子情報通信学会技術研究報告, 115巻, 118(CNR2015 1-5)号, 掲載ページ 11-14, 日本語, 0913-5685, 201502214508579748
  • 人狼における強化学習を用いたエージェントの設計
    梶原 健吾; 鳥海 不二夫; 稲葉 通将
    人工知能学会, 出版日 2015年, 人工知能学会全国大会論文集, 29巻, 掲載ページ 1-3, 日本語, 1347-9881, 40020492003, AA11578981
  • プレイログから獲得した行動選択確率を用いた人狼ゲームのシミュレーション
    平田 佑也; 稲葉 通将; 高橋 健一
    人工知能学会, 出版日 2015年, 人工知能学会全国大会論文集, 29巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020491995, AA11578981
  • 対話エージェントが話し手役になるためのオープンドメイン独話生成
    稲葉 通将; 吉野 友香; 高橋 健一
    人工知能学会, 出版日 2015年, 人工知能学会全国大会論文集, 29巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020496216, AA11578981
  • Twitterから獲得した発話を用いる対話エージェントの構築
    小新田 亮人; 稲葉 通将; 高橋 健一
    人工知能学会, 出版日 2015年, 人工知能学会全国大会論文集, 29巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020496285, AA11578981
  • 人狼知能エージェントによる議論の雰囲気生成とAI識別実験
    片上 大輔; 小林 優; 鳥海 不二夫; 大澤 博隆; 稲葉 通将; 篠田 孝佑
    擬人化エージェントとともにコミュニケーションゲームである人狼を対面環境で行なうことによって高度な対話技術を修得することを目標とし,人間が人狼をプレイする際の議論を実現できる擬人化エージェントを用いた人狼知能対戦システムを開発した. 本研究では,人狼知能対戦システムを用いて人間と人工知能の識別実験を行った.実験の結果,対戦に参加した各エージェントのCMCにおける印象評定尺度の評価には,実験参加者の味方の陣営のエージェントに対して人間と人工知能の間にわずかに差があり,味方陣営の人間と人工知能の見分けが付いている傾向が見られた.また,システムによって表出する動作と発言が統一されている環境では,人間らしさについての評価において人間と人工知能の間に差はなかったため,人狼におけるノンバーバル情報の重要性を確認できた., 日本知能情報ファジィ学会, 出版日 2015年, 日本知能情報ファジィ学会 ファジィ システム シンポジウム 講演論文集, 31巻, 0号, 掲載ページ 370-374, 日本語, 1882-0212, 130005488215, AA12165648
  • ソーシャルメディアにおける対話エージェントとユーザのコミュニケーション分析 (第5回対話システムシンポジウムおよび一般)
    稲葉 通将; 高橋 健一
    人工知能学会, 出版日 2014年12月15日, 言語・音声理解と対話処理研究会, 72巻, 掲載ページ 25-30, 日本語, 0918-5682, 40020291206, AN10432166
  • 議論の構造に着目した人狼ゲームの分析
    稲葉 通将; 大澤 博隆; 片上 大輔; 篠田 孝祐; 鳥海 不二夫
    出版日 2014年10月31日, ゲームプログラミングワークショップ2014論文集, 2014巻, 掲載ページ 61-66, 日本語, 170000087218
  • 人狼知能達成におけるエージェントの推論モデル
    大澤 博隆; 鳥海 不二夫; 稲葉 通将; 片上 大輔; 梶原 健吾; 篠田 孝祐
    出版日 2014年10月31日, ゲームプログラミングワークショップ2014論文集, 2014巻, 掲載ページ 157-161, 日本語, 170000087234
  • 人狼知能サーバの構築
    鳥海 不二夫; 梶原 健吾; 大澤 博隆; 稲葉 通将; 片上 大輔; 篠田 孝祐
    出版日 2014年10月31日, ゲームプログラミングワークショップ2014論文集, 2014巻, 掲載ページ 127-132, 日本語, 170000087228
  • FAN-14-016 人工知能標準問題としての人狼ゲームの提案(OS:人狼知能研究)
    篠田 孝祐; 鳥海 不二夫; 稲葉 通将; 大澤 博隆; 片上 大輔
    一般社団法人日本機械学会, 出版日 2014年09月18日, インテリジェントシステム・シンポジウム講演論文集, 2014巻, 24号, 掲載ページ 74-77, 日本語, 110009964463, AA1190206X
  • FAN-14-017 人狼ゲームのプロトコル設計 : 推理と説得のプロコトル(OS:人狼知能研究)
    大澤 博隆; 鳥海 不二夫; 片上 大輔; 篠田 孝祐; 稲葉 通将
    一般社団法人日本機械学会, 出版日 2014年09月18日, インテリジェントシステム・シンポジウム講演論文集, 2014巻, 24号, 掲載ページ 78-81, 日本語, 110009964464, AA1190206X
  • FAN-14-018 人狼ゲームにおける役職の強さの定量化(OS:人狼知能研究)
    稲葉 通将; 鳥海 不二夫; 大澤 博隆; 片上 大輔; 篠田 孝祐
    一般社団法人日本機械学会, 出版日 2014年09月18日, インテリジェントシステム・シンポジウム講演論文集, 2014巻, 24号, 掲載ページ 82-85, 日本語, 110009964465, AA1190206X
  • FAN-14-019 人狼における学習(OS:人狼知能研究)
    鳥海 不二夫; 稲葉 通将; 大澤 博隆; 片上 大輔; 篠田 孝祐
    一般社団法人日本機械学会, 出版日 2014年09月18日, インテリジェントシステム・シンポジウム講演論文集, 2014巻, 24号, 掲載ページ 86-91, 日本語, 110009964466, AA1190206X
  • 強化学習を用いた人狼における最適戦略の抽出
    梶原健吾; 鳥海不二夫; 大橋弘忠; 大澤博隆; 片上大輔; 稲葉通将; 篠田孝祐; 西野順二
    近年の人工知能の技術の発展により、チェスや将棋等の完全情報ゲームでは人工知能が人間のプロを相手に勝利するようになった。一方で、不完全情報ゲームについての研究は多くはない。そこでコミュニケーションゲーム"人狼"に着目し、人工知能が目指すべき新たな目標として設定する。本研究ではこのゲーム内で使用される主な言葉をプロトコル化し、各エージェントがそのプロトコルで会話をすることでゲームが進行される。過去の会話内容やプレイヤー数等を環境とし、発言内容や能力者の行動等の戦略をQ学習により学習させ、最適戦略を解析する。, 出版日 2014年03月11日, 第76回全国大会講演論文集, 2014巻, 1号, 掲載ページ 597-598, 日本語, 170000085948, AN00349328
  • 学習の休憩時間におけるゲームの効用に関する基礎実験
    竹植 正和; 髙橋 健一; 稲葉 通将
    e -ラーニングでは,学習場所の制約を比較的受けずに自分のペースで学習を行うことができる.一方,学習を進めるうちに次第に飽きてくる傾向がある.本研究では,学習の合間に気分をリフレッシュさせる方法のひとつとして,休憩時間に楽しさを取りいれたゲームを行うことを取り上げ,実験によりその効用を調べる.実験においては.休憩時間に比較的忙しいゲーム,ゆったりとしたゲームをするグループと何もしないグループに学習者を分けて,数式を解く問題を被験者に行わせる.これらのグループの実験結果とアンケートによりこれらのゲームの影響について考察する.In e-learning, learners can study at their own pace and at any place where an internet connection is available. However, users often lose their motivation for learning during the process of study. In this research, we employ games as one of ways to refresh learners during learning. Experiments are conducted to evaluate the effect of playing games during the rest time between learning. In the experiments, participants are separated into three groups: one group plays a busy game, one plays a calm game, and one plays nothing in the rest time. Then, they are asked to solve numerical problems as many as possible in a predetermined time. Using these groups' experimental results and questionnaires, we assess effect of these games., 一般社団法人情報処理学会, 出版日 2014年03月07日, 研究報告コンピュータと教育(CE), 2014巻, 17号, 掲載ページ 1-8, 日本語, 110009684400, AN10096193
  • 同調と反駁に着目した人狼ゲームの分析
    稲葉 通将; 鳥海 不二夫; 大澤 博隆
    人工知能学会, 出版日 2014年, 人工知能学会全国大会論文集, 28巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020075880, AA11578981
  • 人狼ゲームにおける発話行為タグの自動付与
    平田 佑也; 稲葉 通将; 髙橋 健一
    IEEE SMC Hiroshima Chapter, 出版日 2014年, IEEE SMC Hiroshima Chapter Young Researchers' Workshop proceedings = IEEE SMC Hiroshima Chapter若手研究会講演論文集, 掲載ページ 1-4, 日本語, 2187-3577, 40020189211
  • 学生の文理選択に関するアンケートからの知識獲得(言語グリッドと異文化コラボレーション)
    野津田 雄太; 高橋 健一; 稲葉 通将
    近年,国内で理系離れが問題視されている.しかし,理系離れの要因として明確なものは存在しない.そこで本研究では高校生・大学生を対象にアンケートを実施し,データマイニング手法を用いて,学生の生活習慣,科目の履修・嗜好等の傾向から文理選択に影響を与える要因について考察を行う., 一般社団法人電子情報通信学会, 出版日 2013年02月11日, 電子情報通信学会技術研究報告. AI, 人工知能と知識処理, 112巻, 435号, 掲載ページ 17-22, 日本語, 0913-5685, 110009728493, AN10013061
  • 複数の木構造を持つ遺伝的プログラミングの改良
    伊藤 崇; 髙橋 健一; 稲葉 通将
    IEEE SMC Hiroshima Chapter, 出版日 2013年, IEEE SMC Hiroshima Chapter Young Researchers' Workshop proceedings = IEEE SMC Hiroshima Chapter若手研究会講演論文集, 掲載ページ 9-12, 日本語, 2187-3577, 40019881990
  • Twitterを用いた非タスク指向型対話システムの発話生成
    稲葉 通将; 神園 彩香; 高橋 健一
    人工知能学会, 出版日 2013年, 人工知能学会全国大会論文集, 27巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020290820, AA11578981
  • 人狼ゲームデータの統計的分析
    稲葉 通将; 鳥海 不二夫; 高橋 健一
    出版日 2012年11月09日, ゲームプログラミングワークショップ2012論文集, 2012巻, 6号, 掲載ページ 144-147, 日本語, 170000076265
  • 統計的応答手法と話題別学習を用いた非タスク指向型対話エージェント (人工知能学会全国大会(第26回)文化,科学技術と未来) -- (オーガナイズドセッション「OS-03 人を動かすHAI」)
    稲葉 通将; 平井 尚樹; 鳥海 不二夫
    人工知能学会, 出版日 2012年, 人工知能学会全国大会論文集, 26巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020294668, AA11578981
  • 非タスク指向型対話エージェントのためのランキング学習を用いた統計的応答手法 (第二回対話システムシンポジウム 一般セッションと対話システム)
    稲葉 通将; 平井 尚樹; 鳥海 不二夫
    人工知能学会, 出版日 2011年10月13日, 言語・音声理解と対話処理研究会, 63巻, 掲載ページ 25-30, 日本語, 0918-5682, 40019024820, AN10432166
  • SSCI 2011
    稲葉 通将
    出版日 2011年07月01日, 人工知能学会誌 = Journal of Japanese Society for Artificial Intelligence, 26巻, 4号, 掲載ページ 421-421, 日本語, 0912-8085, 10029969494, AN10067140
  • 対話からの話者のプロフィール情報自動獲得
    稲葉 通将; 鳥海 不二夫; 石井 健一郎
    人工知能学会, 出版日 2011年, 人工知能学会全国大会論文集, 25巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020260140, AA11578981
  • 対話エージェントのための発話候補絞り込み
    岩田 直之; 平井 尚樹; 稲葉 通将
    人工知能学会, 出版日 2011年, 人工知能学会全国大会論文集, 25巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020261811, AA11578981
  • 非タスク指向型対話における盛り上がり箇所の抽出
    稲葉 通将; 鳥海 不二夫; 石井 健一郎
    人工知能学会, 出版日 2010年, 人工知能学会全国大会論文集, 24巻, 掲載ページ 1-4, 日本語, 1347-9881, 40020265522, AA11578981
  • 語の共起情報を用いた非タスク指向型対話エージェントの評価
    稲葉 通将; 鳥海 不二夫; 石井 健一郎
    近年,人間と会話のできる対話エージェントが社会に広がりつつある.しかし,人間と雑談をし,話を盛り上げるための非タスク指向型対話エージェントの研究は少なく,知的な応答が可能な非タスク指向型対話エージェントは皆無である.このようなエージェントを実現するためには,客観的・主観的な評価法が不可欠である.本論文では,ある発話が周囲の文脈との間に語彙的結束性を有する場合,その発話は意味的に正しい発話であるという仮定に基づき,対話の意味に踏み込んだ評価手法を提案する.実験の結果,複数の非タスク指向型対話エージェントの性能を比較評価できることが確認された.また,人間による評価との相関が確認された.Recently, computerized dialogue systems have been actively investigated in various fields. However, There are few researches on the systems to talk with a human, and there are no systems that can make an intellectual reply. In order to realize a practical system, the performance should be evaluated quantitatively. In this paper, we propose a new method that is based on the presumption that a utterance is semantically correct if it has lexical cohesion among a context. It was clarified that a our method can evaluate performance of dialogue system objectively and quantitatively., 情報処理学会, 出版日 2009年09月03日, 研究報告数理モデル化と問題解決(MPS), 2009巻, 18号, 掲載ページ 1-6, 日本語, 0919-6072, 110007993834, AN10505667
  • 意味ネットワークによる非タスク指向型対話システムの評価 (「社会システムと知能」および一般発表)
    稲葉 通将; 磯村 直樹; 鳥海 不二夫
    人工知能学会, 出版日 2009年03月02日, 知識ベ-スシステム研究会, 85巻, 掲載ページ 41-46, 日本語, 40016490655, AN10231834
  • 意味ネットワークによる非タスク指向型対話システムの評価(社会システムと知能)
    稲葉 通将; 磯村 直樹; 鳥海 不二夫; 石井 健一郎
    近年,人間と会話のできる対話システムが社会に広がりつつある.しかし,人間と雑談をし,話を盛り上げるための非タスク指向型対話システムの研究は少なく,知的な応答が可能な非タスク指向型対話システムは皆無である.本論文では知的な応答が可能な非タスク指向型対話システムに必要となる知識データ構築を目指し,意味ネットワークを対話から構築する.さらにその応用として非タスク指向型対話システムの客観的・定量的な評価法を提案する.実験の結果,複数の非タスク指向型対話システムの性能を比較評価できることが確認された.また,人間同士の対話についても比較評価できる可能性が確認された., 一般社団法人電子情報通信学会, 出版日 2009年02月23日, 電子情報通信学会技術研究報告. AI, 人工知能と知識処理, 108巻, 456号, 掲載ページ 29-34, 日本語, 0913-5685, 110007325714, AN10013061
  • 意味ネットワークによる非タスク指向型対話システムの評価
    稲葉 通将; 磯村 直樹; 鳥海 不二夫; 石井 健一郎
    近年,人間と会話のできる対話システムが社会に広がりつつある.しかし,人間と雑談をし,話を盛り上げるための非タスク指向型対話システムの研究は少なく,知的な応答が可能な非タスク指向型対話システムは皆無である.本論文では知的な応答が可能な非タスク指向型対話システムに必要となる知識データ構築を目指し,意味ネットワークを対話から構築する.さらにその応用として非タスク指向型対話システムの客観的・定量的な評価法を提案する.実験の結果,複数の非タスク指向型対話システムの性能を比較評価できることが確認された.また,人間同士の対話についても比較評価できる可能性が確認された.Recently, computerized dialogue systems have been actively investigated and used in various fields. However, There are few researches of the systems to talk with a human, and there are no systems that can make an intellectual reply. In this paper, we build a semantic network from a dialogue aiming at the knowledge data construction that is necessary for non-task-oriented dialogue systems. Furthermore, we propose a new criterion which can evaluate non-task-oriented dialogue systems objectively and quantitatively using semantic network. As a result of an experiment, it was clarified that our method can evaluate dialogue systems. In addition, we also carried out a similar experiment with human-human dialogues. The result suggested the possibility that our method can evaluate the dialogues using semantic network., 一般社団法人情報処理学会, 出版日 2009年02月23日, 研究報告知能と複雑系(ICS), 2009巻, 16号, 掲載ページ 29-34, 日本語, 0919-6072, 110007160844, AA11135936

書籍等出版物

  • 深層学習
    柳井, 啓司; 中鹿, 亘; 稲葉, 通将
    日本語, x, 276p, オーム社, 出版日 2022年11月, ISBN 9784274228889
  • Pythonでつくる対話システム
    東中, 竜一郎; 稲葉, 通将; 水上, 雅博
    日本語, x, 246p, オーム社, 出版日 2020年02月, ISBN 9784274224799
  • 人狼知能 : だます・見破る・説得する人工知能
    鳥海, 不二夫; 片上, 大輔; 大澤, 博隆; 稲葉, 通将; 篠田, 孝祐; 狩野, 芳伸
    日本語, vii, 152p, 森北出版, 出版日 2016年08月, ISBN 9784627853713

講演・口頭発表等

  • 人狼知能プロジェクト2015
    篠田 孝祐; 片上 大輔; 稲葉 通将
    日本語, 言語・音声理解と対話処理研究会, http://ci.nii.ac.jp/naid/40020633111
    発表日 2015年10月29日
  • 人狼人工知能プロジェクトの概要と展望
    鳥海不二夫; 稲葉通将; 大澤博隆; 片上大輔; 篠田孝祐; 松原仁
    日本語, 研究報告ゲーム情報学(GI), http://ci.nii.ac.jp/naid/110009799795, 不完全情報コミュニケーションゲームである「人狼」を題材にした研究プロジェクトの概要を紹介し,ゲーム情報学という観点からこのプロジェクトの将来について展望していく.
    発表日 2014年06月28日

所属学協会

  • 情報処理学会
  • 言語処理学会
  • 人工知能学会

共同研究・競争的資金等の研究課題

  • 非対話形式のテキストを用いた対話形式コンテンツの生成
    稲葉 通将
    日本学術振興会, 科学研究費助成事業, 電気通信大学, 挑戦的研究(萌芽), 情報伝達の形式として,対話形式を採用することは古今東西で行われてきた.例えば,古代ギリシャの哲学者プラトンの多くの著作は対話形式で書かれており,また,論語は孔子と弟子の間の対話形式となっている.最近でも,ネット上では対話形式で書かれた記事は多く書かれているほか,漫画を用いたもの(まんがでわかるシリーズ ほか),動画投稿サイトにおける「ゆっくり解説」など,その活用は様々なメディアに広がリ続けている. 対話形式の利点は話し言葉であるため,理解しやすくなること,質問応答の形をとりやすく論理展開が明確になること,登場人物(特に質問者・学習者側)に自己を投影して読むことができること,等が挙げられる.しかし,特にテキストに関しては非対話形式で記述されたものがほとんどである.もちろん,対話形式は網羅的な記述に向かないことや,非対話形式と比較して文量が多くなるといった欠点も存在する.しかし,理解しやすさという観点から,文量が多いものや難解な非対話形式の文書を理解するための導入として有用である.そこで本研究では,非対話形式のテキストから,対話形式のテキストコンテンツを生成する技術開発を実施してきた. 本年度は,非対話形式のデータとして日本語のニュースデータ,対話形式のデータを用いて,そのテーマに関する知識を与えるための対話形式のテキストコンテンツを生成することを目指し,モデル構築のための予備的な検討を実施した., 21K19819
    研究期間 2021年07月09日 - 2024年03月31日
  • モジュール連動に基づく対話システム基盤技術の構築
    東中 竜一郎; 駒谷 和範; 宮尾 祐介; 稲葉 通将; 港 隆史; 槇原 靖
    日本学術振興会, 科学研究費助成事業, 名古屋大学, 新学術領域研究(研究領域提案型), 本研究課題は,現状の対話システムの基本構成を見直し,複数のモジュールが連動することでユーザや状況に応じてシステム全体の効用を最大化することを目的とする.具体的には,「課題A. 他モジュールと連動した音声処理・マルチモーダル処理技術の確立」,「課題B. 他モジュールと連動した言語理解・言語生成技術の確立」,「課題C. 対話の効用に基づく複数モジュールのパラメタ最適化技術」,「課題D. 複数モジュール間での連動プロトコルの設計およびシステム構築とその実証」の4つの課題に取り組む. 課題Aについては,遠方から対話ロボットに近づいてくる話者の性別と年齢を,その歩き姿からリアルタイムに推定する技術を開発した.また,システムがマルチモーダル情報から言外の意図や態度,感情を読み取れるようにするためのデータセットとして,Hazumiを構築し,公開した. 課題Bについては,セマンティックパージングを対話システムに応用するためのデータセットの構築を進めた.自然言語の特性にフォーカスし,対話的図形編集,および,コード生成を対象とした課題設計とデータ構築に着手した. 課題Cについては,パラメタ最適化技術を推進するための対話データ収集を行った.旅行代理店課題を対象とし,ユーザや対話状況に適応的に振舞う対話システムの実現のため,子ども,若年層,高齢者を含む多様な話者のデータを大規模に収集した.加えて,モジュール連動に基づく対話システムの実現手法として,システムの内部パラメタをタスク達成率が最大になるように強化学習を行う手法を提案した. 課題Dについては,モジュール連動プラットフォームに基づく対話システムの実装を進めたほか,コンペティション形式での実証実験を進めた.体験者,開発者,ロボットの3拠点を接続して対話実験を可能とするリモートシステムを開発し,コロナ禍にあっても,予備予選会を滞りなく実施した., 19H05692
    研究期間 2019年06月28日 - 2024年03月31日
  • 上手に説得されることで人を楽しませるゲームAIの構築
    稲葉 通将; 狩野 芳伸; 大槻 恭士
    日本学術振興会, 科学研究費助成事業, 挑戦的研究(萌芽), プレイヤー間で会話することで進行するパーティゲーム「人狼ゲーム」を題材とし,自然言語によりユーザに上手に説得されることでユーザを楽しませるゲームAIの構築手法を提案した. 構築したゲームAIはゲームの状況とユーザとの対話の流れに応じて適切な応答を複数の応答候補から選択することで説得対話を進めるものであり,そのためのマルチタスク学習に基づくニューラルネットワークモデルを提案した.また,自然な説得対話の実現には,AI側はユーザの説得に徐々に応じるように態度を少しずつ変化させる必要がある.そこで,自然な説得の受諾を実現するためのデータドリブンな説得対話管理手法についても提案した., 18K19843
    研究期間 2018年06月29日 - 2021年03月31日
  • 対話的アプローチによる価値観の獲得に関する研究
    稲葉 通将
    日本学術振興会, 科学研究費助成事業, 若手研究(A), 本研究では,(1)大規模Webデータから自動生成した発話候補から,文脈に応じて適切な発話を選択することで対話を進めるリカレントニューラルネットワークを用いた対話モデルの提案,および(2)雑談対話中の発話を入力とし,トピック(例えば音楽,ファッション,健康など)に対するユーザの興味の度合いを推定するモデルの提案を行った・, 16H05880
    研究期間 2016年04月01日 - 2020年03月31日
  • 人狼ゲーム研究:他者中の自己モデルを考えた説得的人工知能の開発に向けて
    稲葉 通将; 鳥海 不二夫; 片上 大輔; 大澤 博隆; 篠田 孝祐
    日本学術振興会, 科学研究費助成事業, 広島市立大学, 挑戦的萌芽研究, 本研究の最終目的は,人と一緒に人狼ゲームをプレイできるエージェントの実現にある.研究期間内にはその第一歩として,人狼ゲーム専用のプロトコルを設計し,それを用いてコミュニケーションを行うエージェントに関する研究を実施した.特に,自分から他者がどう見えるかという「自己の心理モデル」に加え,他者から見て自分がどう見えるかという「他者から見た自己のモデル化」の2つの階層的な心理モデルを持つエージェントを構築した. また,ゲーム研究においては,多くの人が参加して競い合うことで,集合知を集められる環境の整備が重要である.そこで研究開発基盤を整備し,一般参加可能なコンピュータ人狼大会を開催した., 15K12180
    研究期間 2015年04月01日 - 2018年03月31日
  • 学習者の集中度対応型eラーニングインターフェース
    高橋 健一; 稲葉 通将
    日本学術振興会, 科学研究費助成事業, 広島市立大学, 基盤研究(C), 本研究では,コンピュータを用いたeラーニング学習において,学習者の集中を維持するためのインターフェースについて実験を行った.学習者が楽しく学習できるよう,演習問題の正誤の回数により,褒める・叱る画像を表示するシステムを構築し,その有効性を検証した.また,eラーニングが授業などで用いられる場合を想定し,webカメラにより学習者の顔を検出するなどの方法で,真面目に学習しているか否かを検出し,警告を出すシステムを構築し,効果を確かめた.さらに,学習の合間の休憩時間における過ごし方(ゲーム,読書,音楽など)が学習効果に与える影響を調べ,適した過ごし方を明らかにした., 24501141
    研究期間 2012年04月01日 - 2015年03月31日