Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism

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

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

    1 Citation (Scopus)

    Abstract

    This study shows that a novel type of recurrent neural network model can learn to reproduce fluctuating training sequences by inferring their stochastic structures. The network learns to predict not only the mean of the next input state, but also its time-varying variance. The network is trained through maximum likelihood estimation by utilizing the gradient descent method, and the likelihood function is expressed as a function of both the predicted mean and variance. In a numerical experiment, in order to evaluate the performance of the model, we first tested its ability to reproduce fluctuating training sequences generated by a known dynamical system that were perturbed by Gaussian noise with state-dependent variance. Our analysis showed that the network can reproduce the sequences by predicting the variance correctly. Furthermore, the other experiment showed that a humanoid robot equipped with the network can learn to reproduce fluctuating tutoring sequences by inferring latent stochastic structures hidden in the sequences.

    Original languageEnglish
    Title of host publication2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings
    DOIs
    Publication statusPublished - 2013
    Event2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Osaka
    Duration: 2013 Aug 182013 Aug 22

    Other

    Other2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013
    CityOsaka
    Period13/8/1813/8/22

    Fingerprint

    Neural networks
    Recurrent neural networks
    Maximum likelihood estimation
    Dynamical systems
    Experiments
    Robots

    ASJC Scopus subject areas

    • Artificial Intelligence
    • Human-Computer Interaction
    • Software

    Cite this

    Murata, S., Namikawa, J., Arie, H., Tani, J., & Sugano, S. (2013). Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism. In 2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings [6652545] https://doi.org/10.1109/DevLrn.2013.6652545

    Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism. / Murata, Shingo; Namikawa, Jun; Arie, Hiroaki; Tani, Jun; Sugano, Shigeki.

    2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings. 2013. 6652545.

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

    Murata, S, Namikawa, J, Arie, H, Tani, J & Sugano, S 2013, Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism. in 2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings., 6652545, 2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013, Osaka, 13/8/18. https://doi.org/10.1109/DevLrn.2013.6652545
    Murata S, Namikawa J, Arie H, Tani J, Sugano S. Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism. In 2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings. 2013. 6652545 https://doi.org/10.1109/DevLrn.2013.6652545
    Murata, Shingo ; Namikawa, Jun ; Arie, Hiroaki ; Tani, Jun ; Sugano, Shigeki. / Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism. 2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings. 2013.
    @inproceedings{470f06bc20ed4b0aa56d78fbecdee2ad,
    title = "Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism",
    abstract = "This study shows that a novel type of recurrent neural network model can learn to reproduce fluctuating training sequences by inferring their stochastic structures. The network learns to predict not only the mean of the next input state, but also its time-varying variance. The network is trained through maximum likelihood estimation by utilizing the gradient descent method, and the likelihood function is expressed as a function of both the predicted mean and variance. In a numerical experiment, in order to evaluate the performance of the model, we first tested its ability to reproduce fluctuating training sequences generated by a known dynamical system that were perturbed by Gaussian noise with state-dependent variance. Our analysis showed that the network can reproduce the sequences by predicting the variance correctly. Furthermore, the other experiment showed that a humanoid robot equipped with the network can learn to reproduce fluctuating tutoring sequences by inferring latent stochastic structures hidden in the sequences.",
    author = "Shingo Murata and Jun Namikawa and Hiroaki Arie and Jun Tani and Shigeki Sugano",
    year = "2013",
    doi = "10.1109/DevLrn.2013.6652545",
    language = "English",
    isbn = "9781479910366",
    booktitle = "2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings",

    }

    TY - GEN

    T1 - Learning to reproduce fluctuating behavioral sequences using a dynamic neural network model with time-varying variance estimation mechanism

    AU - Murata, Shingo

    AU - Namikawa, Jun

    AU - Arie, Hiroaki

    AU - Tani, Jun

    AU - Sugano, Shigeki

    PY - 2013

    Y1 - 2013

    N2 - This study shows that a novel type of recurrent neural network model can learn to reproduce fluctuating training sequences by inferring their stochastic structures. The network learns to predict not only the mean of the next input state, but also its time-varying variance. The network is trained through maximum likelihood estimation by utilizing the gradient descent method, and the likelihood function is expressed as a function of both the predicted mean and variance. In a numerical experiment, in order to evaluate the performance of the model, we first tested its ability to reproduce fluctuating training sequences generated by a known dynamical system that were perturbed by Gaussian noise with state-dependent variance. Our analysis showed that the network can reproduce the sequences by predicting the variance correctly. Furthermore, the other experiment showed that a humanoid robot equipped with the network can learn to reproduce fluctuating tutoring sequences by inferring latent stochastic structures hidden in the sequences.

    AB - This study shows that a novel type of recurrent neural network model can learn to reproduce fluctuating training sequences by inferring their stochastic structures. The network learns to predict not only the mean of the next input state, but also its time-varying variance. The network is trained through maximum likelihood estimation by utilizing the gradient descent method, and the likelihood function is expressed as a function of both the predicted mean and variance. In a numerical experiment, in order to evaluate the performance of the model, we first tested its ability to reproduce fluctuating training sequences generated by a known dynamical system that were perturbed by Gaussian noise with state-dependent variance. Our analysis showed that the network can reproduce the sequences by predicting the variance correctly. Furthermore, the other experiment showed that a humanoid robot equipped with the network can learn to reproduce fluctuating tutoring sequences by inferring latent stochastic structures hidden in the sequences.

    UR - http://www.scopus.com/inward/record.url?scp=84891133751&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84891133751&partnerID=8YFLogxK

    U2 - 10.1109/DevLrn.2013.6652545

    DO - 10.1109/DevLrn.2013.6652545

    M3 - Conference contribution

    AN - SCOPUS:84891133751

    SN - 9781479910366

    BT - 2013 IEEE 3rd Joint International Conference on Development and Learning and Epigenetic Robotics, ICDL 2013 - Electronic Conference Proceedings

    ER -