IDEAS home Printed from https://ideas.repec.org/a/plo/pcbi00/1007263.html
   My bibliography  Save this article

Identifying nonlinear dynamical systems via generative recurrent neural networks with applications to fMRI

Author

Listed:
  • Georgia Koppe
  • Hazem Toutounji
  • Peter Kirsch
  • Stefanie Lis
  • Daniel Durstewitz

Abstract

A major tenet in theoretical neuroscience is that cognitive and behavioral processes are ultimately implemented in terms of the neural system dynamics. Accordingly, a major aim for the analysis of neurophysiological measurements should lie in the identification of the computational dynamics underlying task processing. Here we advance a state space model (SSM) based on generative piecewise-linear recurrent neural networks (PLRNN) to assess dynamics from neuroimaging data. In contrast to many other nonlinear time series models which have been proposed for reconstructing latent dynamics, our model is easily interpretable in neural terms, amenable to systematic dynamical systems analysis of the resulting set of equations, and can straightforwardly be transformed into an equivalent continuous-time dynamical system. The major contributions of this paper are the introduction of a new observation model suitable for functional magnetic resonance imaging (fMRI) coupled to the latent PLRNN, an efficient stepwise training procedure that forces the latent model to capture the ‘true’ underlying dynamics rather than just fitting (or predicting) the observations, and of an empirical measure based on the Kullback-Leibler divergence to evaluate from empirical time series how well this goal of approximating the underlying dynamics has been achieved. We validate and illustrate the power of our approach on simulated ‘ground-truth’ dynamical systems as well as on experimental fMRI time series, and demonstrate that the learnt dynamics harbors task-related nonlinear structure that a linear dynamical model fails to capture. Given that fMRI is one of the most common techniques for measuring brain activity non-invasively in human subjects, this approach may provide a novel step toward analyzing aberrant (nonlinear) dynamics for clinical assessment or neuroscientific research.Author summary: Computational processes in the brain are often assumed to be implemented in terms of nonlinear neural network dynamics. However, experimentally we usually do not have direct access to this underlying dynamical process that generated the observed time series, but have to infer it from a sample of noisy and mixed measurements like fMRI data. Here we combine a dynamically universal recurrent neural network (RNN) model for approximating the unknown system dynamics with an observation model that links this dynamics to experimental measurements, taking fMRI data as an example. We develop a new stepwise optimization algorithm, within the statistical framework of state space models, that forces the latent RNN model toward the true data-generating dynamical process, and demonstrate its power on benchmark systems like the chaotic Lorenz attractor. We also introduce a novel, fast-to-compute measure for assessing how well this worked out in any empirical situation for which the ground truth dynamical system is not known. RNN models trained on human fMRI data this way can generate new data with the same temporal structure and properties, and exhibit interesting nonlinear dynamical phenomena related to experimental task conditions and behavioral performance. This approach can easily be generalized to many other recording modalities.

Suggested Citation

  • Georgia Koppe & Hazem Toutounji & Peter Kirsch & Stefanie Lis & Daniel Durstewitz, 2019. "Identifying nonlinear dynamical systems via generative recurrent neural networks with applications to fMRI," PLOS Computational Biology, Public Library of Science, vol. 15(8), pages 1-35, August.
  • Handle: RePEc:plo:pcbi00:1007263
    DOI: 10.1371/journal.pcbi.1007263
    as

    Download full text from publisher

    File URL: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1007263
    Download Restriction: no

    File URL: https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1007263&type=printable
    Download Restriction: no

    File URL: https://libkey.io/10.1371/journal.pcbi.1007263?utm_source=ideas
    LibKey link: if access is restricted and if your library uses this service, LibKey will redirect you to where you can use your library subscription to access this item
    ---><---

    References listed on IDEAS

    as
    1. Bergmeir, Christoph & Hyndman, Rob J. & Koo, Bonsoo, 2018. "A note on the validity of cross-validation for evaluating autoregressive time series prediction," Computational Statistics & Data Analysis, Elsevier, vol. 120(C), pages 70-83.
    2. Koyama, Shinsuke & Castellanos Pérez-Bolde, Lucia & Shalizi, Cosma Rohilla & Kass, Robert E., 2010. "Approximate Methods for State-Space Models," Journal of the American Statistical Association, American Statistical Association, vol. 105(489), pages 170-180.
    3. Valerio Mante & David Sussillo & Krishna V. Shenoy & William T. Newsome, 2013. "Context-dependent computation by recurrent dynamics in prefrontal cortex," Nature, Nature, vol. 503(7474), pages 78-84, November.
    4. Emili Balaguer-Ballester & Christopher C Lapish & Jeremy K Seamans & Daniel Durstewitz, 2011. "Attracting Dynamics of Frontal Cortex Ensembles during Memory-Guided Decision-Making," PLOS Computational Biology, Public Library of Science, vol. 7(5), pages 1-19, May.
    5. Simon N. Wood, 2010. "Statistical inference for noisy nonlinear ecological dynamic systems," Nature, Nature, vol. 466(7310), pages 1102-1104, August.
    6. Durbin, James & Koopman, Siem Jan, 2012. "Time Series Analysis by State Space Methods," OUP Catalogue, Oxford University Press, edition 2, number 9780199641178.
    7. Daniel Durstewitz, 2017. "A state space approach for piecewise-linear recurrent neural networks for identifying computational dynamics from neural measurements," PLOS Computational Biology, Public Library of Science, vol. 13(6), pages 1-33, June.
    Full references (including those not matched with items on IDEAS)

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. Daniel Durstewitz, 2017. "A state space approach for piecewise-linear recurrent neural networks for identifying computational dynamics from neural measurements," PLOS Computational Biology, Public Library of Science, vol. 13(6), pages 1-33, June.
    2. Paolo Maranzano & Alessandro Fassò & Matteo Pelagatti & Manfred Mudelsee, 2020. "Statistical Modeling of the Early-Stage Impact of a New Traffic Policy in Milan, Italy," IJERPH, MDPI, vol. 17(3), pages 1-22, February.
    3. Petropoulos, Fotios & Apiletti, Daniele & Assimakopoulos, Vassilios & Babai, Mohamed Zied & Barrow, Devon K. & Ben Taieb, Souhaib & Bergmeir, Christoph & Bessa, Ricardo J. & Bijak, Jakub & Boylan, Joh, 2022. "Forecasting: theory and practice," International Journal of Forecasting, Elsevier, vol. 38(3), pages 705-871.
      • Fotios Petropoulos & Daniele Apiletti & Vassilios Assimakopoulos & Mohamed Zied Babai & Devon K. Barrow & Souhaib Ben Taieb & Christoph Bergmeir & Ricardo J. Bessa & Jakub Bijak & John E. Boylan & Jet, 2020. "Forecasting: theory and practice," Papers 2012.03854, arXiv.org, revised Jan 2022.
    4. Michael D. Hunter & Haya Fatimah & Marina A. Bornovalova, 2022. "Two Filtering Methods of Forecasting Linear and Nonlinear Dynamics of Intensive Longitudinal Data," Psychometrika, Springer;The Psychometric Society, vol. 87(2), pages 477-505, June.
    5. Trond Husby & Hans Visser, 2021. "Short- to medium-run forecasting of mobility with dynamic linear models," Demographic Research, Max Planck Institute for Demographic Research, Rostock, Germany, vol. 45(28), pages 871-902.
    6. Avanzi, Benjamin & Taylor, Greg & Vu, Phuong Anh & Wong, Bernard, 2020. "A multivariate evolutionary generalised linear model framework with adaptive estimation for claims reserving," Insurance: Mathematics and Economics, Elsevier, vol. 93(C), pages 50-71.
    7. François R. Velde, 2009. "Chronicle of a Deflation Unforetold," Journal of Political Economy, University of Chicago Press, vol. 117(4), pages 591-634, August.
    8. Wen Xu, 2016. "Estimation of Dynamic Panel Data Models with Stochastic Volatility Using Particle Filters," Econometrics, MDPI, vol. 4(4), pages 1-13, October.
    9. Henriques, Irene & Sadorsky, Perry, 2023. "Forecasting rare earth stock prices with machine learning," Resources Policy, Elsevier, vol. 86(PA).
    10. Alejandro Rodriguez & Esther Ruiz, 2009. "Bootstrap prediction intervals in state–space models," Journal of Time Series Analysis, Wiley Blackwell, vol. 30(2), pages 167-178, March.
    11. Parrini, Alessandro, 2013. "Importance Sampling for Portfolio Credit Risk in Factor Copula Models," MPRA Paper 103745, University Library of Munich, Germany.
    12. Grazzini, Jakob & Richiardi, Matteo G. & Tsionas, Mike, 2017. "Bayesian estimation of agent-based models," Journal of Economic Dynamics and Control, Elsevier, vol. 77(C), pages 26-47.
    13. Jean-Luc Gaffard, 2014. "Crise de la théorie et crise de la politique économique. Des modèles d'équilibre général stochastique aux modèles de dynamique hors de l'équilibre," Revue économique, Presses de Sciences-Po, vol. 65(1), pages 71-96.
    14. Salman Huseynov, 2021. "Long and short memory in dynamic term structure models," CREATES Research Papers 2021-15, Department of Economics and Business Economics, Aarhus University.
    15. Tsionas, Mike G., 2021. "Bayesian forecasting with the structural damped trend model," International Journal of Production Economics, Elsevier, vol. 234(C).
    16. Philippe Goulet Coulombe & Maxime Leroux & Dalibor Stevanovic & Stéphane Surprenant, 2022. "How is machine learning useful for macroeconomic forecasting?," Journal of Applied Econometrics, John Wiley & Sons, Ltd., vol. 37(5), pages 920-964, August.
    17. Tommaso Proietti, 2002. "Some Reflections on Trend-Cycle Decompositions with Correlated Components," Econometrics 0209002, University Library of Munich, Germany.
    18. Tobias Hartl & Roland Jucknewitz, 2022. "Approximate state space modelling of unobserved fractional components," Econometric Reviews, Taylor & Francis Journals, vol. 41(1), pages 75-98, January.
    19. Broto Carmen & Ruiz Esther, 2009. "Testing for Conditional Heteroscedasticity in the Components of Inflation," Studies in Nonlinear Dynamics & Econometrics, De Gruyter, vol. 13(2), pages 1-30, May.
    20. Marczak, Martyna & Proietti, Tommaso, 2016. "Outlier detection in structural time series models: The indicator saturation approach," International Journal of Forecasting, Elsevier, vol. 32(1), pages 180-202.

    More about this item

    Statistics

    Access and download statistics

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:plo:pcbi00:1007263. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: ploscompbiol (email available below). General contact details of provider: https://journals.plos.org/ploscompbiol/ .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.