Development of proactive and reactive behavior via meta-learning of prediction error variance

Shingo Murata, Jun Namikawa, Hiroaki Arie, Jun Tani, Shigeki Sugano

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

    Abstract

    This paper investigates a possible neurodynamic mechanism that enables autonomous switching between two basic behavioral modes, namely a "proactive mode" and a "reactive mode." In the proactive mode, actions are generated as intended, whereas in the reactive mode actions are generated in response to the sensory state.We conducted neurorobotics experiments to investigate how these two modes can develop and how a robot can learn to switch autonomously between the two modes as necessary by utilizing our recently developed dynamic neural network model. Tasks designed for the robot included switching between proactive imitation of other's predictable movements using acquired memories and reactive following of other's unpredictable movements through iterative learning of alternating predictable and unpredictable movement patterns. The experimental results revealed that this "meta-learning" capability can lead to self-organization of adequate contextual dynamical structures that can perform autonomous switching between the different behavioral modes.

    Original languageEnglish
    Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Pages537-544
    Number of pages8
    Volume8226 LNCS
    EditionPART 1
    DOIs
    Publication statusPublished - 2013
    Event20th International Conference on Neural Information Processing, ICONIP 2013 - Daegu
    Duration: 2013 Nov 32013 Nov 7

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    NumberPART 1
    Volume8226 LNCS
    ISSN (Print)03029743
    ISSN (Electronic)16113349

    Other

    Other20th International Conference on Neural Information Processing, ICONIP 2013
    CityDaegu
    Period13/11/313/11/7

    Fingerprint

    Meta-learning
    Prediction Error
    Robot
    Robots
    Dynamic Neural Networks
    Imitation
    Self-organization
    Neural Network Model
    Switch
    Switches
    Neural networks
    Data storage equipment
    Necessary
    Experimental Results
    Experiment
    Movement
    Experiments

    Keywords

    • Humanoid robot
    • Neurorobotics
    • Recurrent neural network

    ASJC Scopus subject areas

    • Computer Science(all)
    • Theoretical Computer Science

    Cite this

    Murata, S., Namikawa, J., Arie, H., Tani, J., & Sugano, S. (2013). Development of proactive and reactive behavior via meta-learning of prediction error variance. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 1 ed., Vol. 8226 LNCS, pp. 537-544). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8226 LNCS, No. PART 1). https://doi.org/10.1007/978-3-642-42054-2_67

    Development of proactive and reactive behavior via meta-learning of prediction error variance. / Murata, Shingo; Namikawa, Jun; Arie, Hiroaki; Tani, Jun; Sugano, Shigeki.

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8226 LNCS PART 1. ed. 2013. p. 537-544 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8226 LNCS, No. PART 1).

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

    Murata, S, Namikawa, J, Arie, H, Tani, J & Sugano, S 2013, Development of proactive and reactive behavior via meta-learning of prediction error variance. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 edn, vol. 8226 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 8226 LNCS, pp. 537-544, 20th International Conference on Neural Information Processing, ICONIP 2013, Daegu, 13/11/3. https://doi.org/10.1007/978-3-642-42054-2_67
    Murata S, Namikawa J, Arie H, Tani J, Sugano S. Development of proactive and reactive behavior via meta-learning of prediction error variance. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 1 ed. Vol. 8226 LNCS. 2013. p. 537-544. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). https://doi.org/10.1007/978-3-642-42054-2_67
    Murata, Shingo ; Namikawa, Jun ; Arie, Hiroaki ; Tani, Jun ; Sugano, Shigeki. / Development of proactive and reactive behavior via meta-learning of prediction error variance. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8226 LNCS PART 1. ed. 2013. pp. 537-544 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
    @inproceedings{5e481c8ddabf41c4a03af6f07da3d6c7,
    title = "Development of proactive and reactive behavior via meta-learning of prediction error variance",
    abstract = "This paper investigates a possible neurodynamic mechanism that enables autonomous switching between two basic behavioral modes, namely a {"}proactive mode{"} and a {"}reactive mode.{"} In the proactive mode, actions are generated as intended, whereas in the reactive mode actions are generated in response to the sensory state.We conducted neurorobotics experiments to investigate how these two modes can develop and how a robot can learn to switch autonomously between the two modes as necessary by utilizing our recently developed dynamic neural network model. Tasks designed for the robot included switching between proactive imitation of other's predictable movements using acquired memories and reactive following of other's unpredictable movements through iterative learning of alternating predictable and unpredictable movement patterns. The experimental results revealed that this {"}meta-learning{"} capability can lead to self-organization of adequate contextual dynamical structures that can perform autonomous switching between the different behavioral modes.",
    keywords = "Humanoid robot, Neurorobotics, Recurrent neural network",
    author = "Shingo Murata and Jun Namikawa and Hiroaki Arie and Jun Tani and Shigeki Sugano",
    year = "2013",
    doi = "10.1007/978-3-642-42054-2_67",
    language = "English",
    isbn = "9783642420535",
    volume = "8226 LNCS",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    number = "PART 1",
    pages = "537--544",
    booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    edition = "PART 1",

    }

    TY - GEN

    T1 - Development of proactive and reactive behavior via meta-learning of prediction error variance

    AU - Murata, Shingo

    AU - Namikawa, Jun

    AU - Arie, Hiroaki

    AU - Tani, Jun

    AU - Sugano, Shigeki

    PY - 2013

    Y1 - 2013

    N2 - This paper investigates a possible neurodynamic mechanism that enables autonomous switching between two basic behavioral modes, namely a "proactive mode" and a "reactive mode." In the proactive mode, actions are generated as intended, whereas in the reactive mode actions are generated in response to the sensory state.We conducted neurorobotics experiments to investigate how these two modes can develop and how a robot can learn to switch autonomously between the two modes as necessary by utilizing our recently developed dynamic neural network model. Tasks designed for the robot included switching between proactive imitation of other's predictable movements using acquired memories and reactive following of other's unpredictable movements through iterative learning of alternating predictable and unpredictable movement patterns. The experimental results revealed that this "meta-learning" capability can lead to self-organization of adequate contextual dynamical structures that can perform autonomous switching between the different behavioral modes.

    AB - This paper investigates a possible neurodynamic mechanism that enables autonomous switching between two basic behavioral modes, namely a "proactive mode" and a "reactive mode." In the proactive mode, actions are generated as intended, whereas in the reactive mode actions are generated in response to the sensory state.We conducted neurorobotics experiments to investigate how these two modes can develop and how a robot can learn to switch autonomously between the two modes as necessary by utilizing our recently developed dynamic neural network model. Tasks designed for the robot included switching between proactive imitation of other's predictable movements using acquired memories and reactive following of other's unpredictable movements through iterative learning of alternating predictable and unpredictable movement patterns. The experimental results revealed that this "meta-learning" capability can lead to self-organization of adequate contextual dynamical structures that can perform autonomous switching between the different behavioral modes.

    KW - Humanoid robot

    KW - Neurorobotics

    KW - Recurrent neural network

    UR - http://www.scopus.com/inward/record.url?scp=84893353213&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84893353213&partnerID=8YFLogxK

    U2 - 10.1007/978-3-642-42054-2_67

    DO - 10.1007/978-3-642-42054-2_67

    M3 - Conference contribution

    SN - 9783642420535

    VL - 8226 LNCS

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

    SP - 537

    EP - 544

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

    ER -