Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN

Tatsuro Yamada, Shingo Murata, Hiroaki Arie, Tetsuya Ogata

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    2 Citations (Scopus)

    Abstract

    Meanings of language expressions are constructed not only from words grounded in real-world matters, but also from words such as “not” that participate in the construction by working as logical operators. This study proposes a connectionist method for learning and internally representing functions that deal with both of these word groups, and grounding sentences constructed from them in corresponding behaviors just by experiencing raw sequential data of an imposed task. In the experiment, a robot implemented with a recurrent neural network is required to ground imperative positive and negative sentences given as a sequence of words in corresponding goal-oriented behavior. Analysis of the internal representations reveals that the network fulfilled the requirement by extracting XOR problems implicitly included in the target sequences and solving them by learning to represent the logical operations in its nonlinear dynamics in a self-organizing manner.

    Original languageEnglish
    Title of host publicationArtificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings
    PublisherSpringer Verlag
    Pages339-346
    Number of pages8
    Volume9886 LNCS
    ISBN (Print)9783319447773
    DOIs
    Publication statusPublished - 2016
    Event25th International Conference on Artificial Neural Networks, ICANN 2016 - Barcelona, Spain
    Duration: 2016 Sep 62016 Sep 9

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume9886 LNCS
    ISSN (Print)03029743
    ISSN (Electronic)16113349

    Other

    Other25th International Conference on Artificial Neural Networks, ICANN 2016
    CountrySpain
    CityBarcelona
    Period16/9/616/9/9

    Fingerprint

    Recurrent neural networks
    Electric grounding
    Linking
    Robot
    Logical operator
    Robots
    Recurrent Neural Networks
    Self-organizing
    Nonlinear Dynamics
    Experiments
    Internal
    Target
    Requirements
    Experiment
    Learning
    Meaning
    Language

    Keywords

    • Human–robot interaction
    • Logical operation
    • Recurrent neural network
    • Symbol grounding

    ASJC Scopus subject areas

    • Theoretical Computer Science
    • Computer Science(all)

    Cite this

    Yamada, T., Murata, S., Arie, H., & Ogata, T. (2016). Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN. In Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings (Vol. 9886 LNCS, pp. 339-346). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9886 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-44778-0_40

    Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN. / Yamada, Tatsuro; Murata, Shingo; Arie, Hiroaki; Ogata, Tetsuya.

    Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Vol. 9886 LNCS Springer Verlag, 2016. p. 339-346 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9886 LNCS).

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Yamada, T, Murata, S, Arie, H & Ogata, T 2016, Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN. in Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. vol. 9886 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9886 LNCS, Springer Verlag, pp. 339-346, 25th International Conference on Artificial Neural Networks, ICANN 2016, Barcelona, Spain, 16/9/6. https://doi.org/10.1007/978-3-319-44778-0_40
    Yamada T, Murata S, Arie H, Ogata T. Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN. In Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Vol. 9886 LNCS. Springer Verlag. 2016. p. 339-346. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-44778-0_40
    Yamada, Tatsuro ; Murata, Shingo ; Arie, Hiroaki ; Ogata, Tetsuya. / Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN. Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Vol. 9886 LNCS Springer Verlag, 2016. pp. 339-346 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{93a6f97de7274de285e4b0f14d0a2604,
    title = "Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN",
    abstract = "Meanings of language expressions are constructed not only from words grounded in real-world matters, but also from words such as “not” that participate in the construction by working as logical operators. This study proposes a connectionist method for learning and internally representing functions that deal with both of these word groups, and grounding sentences constructed from them in corresponding behaviors just by experiencing raw sequential data of an imposed task. In the experiment, a robot implemented with a recurrent neural network is required to ground imperative positive and negative sentences given as a sequence of words in corresponding goal-oriented behavior. Analysis of the internal representations reveals that the network fulfilled the requirement by extracting XOR problems implicitly included in the target sequences and solving them by learning to represent the logical operations in its nonlinear dynamics in a self-organizing manner.",
    keywords = "Human–robot interaction, Logical operation, Recurrent neural network, Symbol grounding",
    author = "Tatsuro Yamada and Shingo Murata and Hiroaki Arie and Tetsuya Ogata",
    year = "2016",
    doi = "10.1007/978-3-319-44778-0_40",
    language = "English",
    isbn = "9783319447773",
    volume = "9886 LNCS",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    publisher = "Springer Verlag",
    pages = "339--346",
    booktitle = "Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings",
    address = "Germany",

    }

    TY - GEN

    T1 - Dynamical linking of positive and negative sentences to goal-oriented robot behavior by hierarchical RNN

    AU - Yamada, Tatsuro

    AU - Murata, Shingo

    AU - Arie, Hiroaki

    AU - Ogata, Tetsuya

    PY - 2016

    Y1 - 2016

    N2 - Meanings of language expressions are constructed not only from words grounded in real-world matters, but also from words such as “not” that participate in the construction by working as logical operators. This study proposes a connectionist method for learning and internally representing functions that deal with both of these word groups, and grounding sentences constructed from them in corresponding behaviors just by experiencing raw sequential data of an imposed task. In the experiment, a robot implemented with a recurrent neural network is required to ground imperative positive and negative sentences given as a sequence of words in corresponding goal-oriented behavior. Analysis of the internal representations reveals that the network fulfilled the requirement by extracting XOR problems implicitly included in the target sequences and solving them by learning to represent the logical operations in its nonlinear dynamics in a self-organizing manner.

    AB - Meanings of language expressions are constructed not only from words grounded in real-world matters, but also from words such as “not” that participate in the construction by working as logical operators. This study proposes a connectionist method for learning and internally representing functions that deal with both of these word groups, and grounding sentences constructed from them in corresponding behaviors just by experiencing raw sequential data of an imposed task. In the experiment, a robot implemented with a recurrent neural network is required to ground imperative positive and negative sentences given as a sequence of words in corresponding goal-oriented behavior. Analysis of the internal representations reveals that the network fulfilled the requirement by extracting XOR problems implicitly included in the target sequences and solving them by learning to represent the logical operations in its nonlinear dynamics in a self-organizing manner.

    KW - Human–robot interaction

    KW - Logical operation

    KW - Recurrent neural network

    KW - Symbol grounding

    UR - http://www.scopus.com/inward/record.url?scp=84987935120&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84987935120&partnerID=8YFLogxK

    U2 - 10.1007/978-3-319-44778-0_40

    DO - 10.1007/978-3-319-44778-0_40

    M3 - Conference contribution

    SN - 9783319447773

    VL - 9886 LNCS

    T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    SP - 339

    EP - 346

    BT - Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings

    PB - Springer Verlag

    ER -