Pith. sign in

REVIEW 2 major objections 2 cited by

Embedding the Hopf normal form in a probabilistic state-space model lets researchers extract natural frequency, Floquet exponent, and asymptotic phase directly from noisy oscillatory time series.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-06-26 01:48 UTC pith:OX6JIMVT

load-bearing objection The paper embeds the Hopf normal form in a state-space model for joint parameter and state inference from noisy oscillations, plus GP-based sensitivity reconstruction, but the far-from-bifurcation robustness rests on an unverified modeling assumption. the 2 major comments →

arxiv 2606.26847 v1 pith:OX6JIMVT submitted 2026-06-25 nlin.AO

Data-driven inference of Hopf normal form representations from oscillatory time series

Shinsuke Koyama , Ryota Kobayashi This is my paper
classification nlin.AO
keywords Hopf normal formoscillatory time seriesstate-space modeldata-driven inferenceFloquet exponentasymptotic phasephase sensitivityvan der Pol oscillator
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents a method to fit the Hopf normal form to observed time series without prior knowledge of the governing equations. It places the normal form inside a probabilistic state-space model so that latent states and model parameters are inferred together from the data. This joint inference produces stable estimates of frequency, stability measures, and phase even when noise is strong or the system sits well away from the onset of oscillation. The same framework, paired with complex Gaussian process regression, also recovers the system's phase and amplitude sensitivity functions. Benchmarks against the van der Pol oscillator show clear gains in accuracy over earlier phase-based and regression techniques.

Core claim

We introduce a data-driven framework that maps noisy oscillatory time series directly onto the Hopf normal form, enabling inference of underlying dynamics without knowledge of governing equations. By embedding the normal form in a probabilistic state-space model, the method jointly infers latent states and system parameters, yielding robust estimates of the natural frequency, Floquet exponent, and asymptotic phase even far from the bifurcation point and under strong noise. Combined with complex Gaussian process regression, the approach further reconstructs phase and amplitude sensitivity functions from data.

What carries the argument

Probabilistic state-space model embedding of the Hopf normal form, which performs joint inference of latent states and parameters from data.

Load-bearing premise

The observed oscillatory dynamics are sufficiently well approximated by the Hopf normal form inside the chosen probabilistic state-space model.

What would settle it

Applying the method to data generated from a system whose dynamics are known to lie outside the Hopf normal form (for example, a chaotic oscillator) produces inconsistent or unstable parameter estimates and poor reconstruction of the time series.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Robust estimates of natural frequency, Floquet exponent, and asymptotic phase from data alone.
  • Reconstruction of phase and amplitude sensitivity functions via complex Gaussian process regression.
  • Substantially improved accuracy and noise robustness relative to existing phase-based and regression methods.
  • A general route to low-dimensional descriptions of oscillatory dynamics in complex systems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The approach could be tested on experimental recordings from biological or engineered oscillators to see whether the inferred parameters predict observed stability changes.
  • It may be possible to replace the Hopf normal form with other normal forms inside the same state-space structure for systems near different bifurcations.
  • The inferred sensitivity functions could be used to design control inputs that shift the oscillation frequency or amplitude in a data-driven manner.
  • If the method remains accurate on real-world data with unknown noise statistics, it would reduce the need for detailed mechanistic models when only rhythmic behavior is of interest.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 0 minor

Summary. The manuscript introduces a data-driven framework that embeds the cubic Hopf normal form in a probabilistic state-space model to jointly infer latent states and system parameters (natural frequency, Floquet exponent, asymptotic phase) directly from noisy oscillatory time series. It further combines the approach with complex Gaussian process regression to reconstruct phase and amplitude sensitivity functions, and reports substantially improved accuracy and noise robustness on van der Pol oscillator benchmarks relative to existing phase-based and regression methods, with the central claim being that the estimates remain reliable even far from the bifurcation point.

Significance. If the inferred quantities can be shown to match independent ground-truth values of the underlying system rather than effective parameters of the truncated model, the work would provide a practical bridge between normal-form theory and statistical inference, enabling low-dimensional dynamical descriptions of oscillatory data in complex systems without requiring the full governing equations.

major comments (2)
  1. [Abstract] Abstract: the central claim that the method yields 'robust estimates ... even far from the bifurcation point' is load-bearing and requires explicit verification that the inferred natural frequency, Floquet exponent, and asymptotic phase match quantities computed independently from the original system (e.g., analytic or numerical values for the van der Pol oscillator at large μ) rather than being artifacts of enforcing the cubic truncation inside the state-space model.
  2. The manuscript does not appear to include controlled tests in which the Hopf normal-form approximation is deliberately violated (e.g., by increasing μ well beyond the regime where higher-order terms remain negligible) to quantify when the joint inference procedure fails to recover the true dynamical quantities.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We agree that the central claims require stronger validation against independent ground truth and that the limits of the normal-form approximation should be quantified. We will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the method yields 'robust estimates ... even far from the bifurcation point' is load-bearing and requires explicit verification that the inferred natural frequency, Floquet exponent, and asymptotic phase match quantities computed independently from the original system (e.g., analytic or numerical values for the van der Pol oscillator at large μ) rather than being artifacts of enforcing the cubic truncation inside the state-space model.

    Authors: We acknowledge that direct comparison to independent ground-truth values is necessary to substantiate the claim. In the revised manuscript we will add explicit side-by-side comparisons for the van der Pol oscillator, using known analytic expressions for natural frequency and Floquet exponent together with numerical computation of asymptotic phase, across a range of μ values including those far from the bifurcation. This will demonstrate whether the inferred quantities recover the true system parameters or are artifacts of the cubic truncation. revision: yes

  2. Referee: The manuscript does not appear to include controlled tests in which the Hopf normal-form approximation is deliberately violated (e.g., by increasing μ well beyond the regime where higher-order terms remain negligible) to quantify when the joint inference procedure fails to recover the true dynamical quantities.

    Authors: We agree that controlled violation tests are required to delineate the method's applicability. The revision will include additional experiments on the van der Pol oscillator with progressively larger μ (e.g., μ = 5, 10, 20) where higher-order terms become significant; inferred parameters will be compared to independent ground-truth values to quantify the deviation and identify the regime in which the cubic model ceases to recover true quantities. revision: yes

Circularity Check

0 steps flagged

No circularity: inference outputs are fitted parameters, not self-defined reductions

full rationale

The paper describes embedding the Hopf normal form inside a probabilistic state-space model to jointly infer latent states and parameters (natural frequency, Floquet exponent, asymptotic phase) directly from noisy time series. These quantities are the explicit parameters of the embedded model; their estimation is the intended output of the fitting procedure rather than a derived claim that reduces by construction to the inputs. No equations or steps are shown that rename a fitted value as an independent prediction, invoke self-citation for a uniqueness theorem, or smuggle an ansatz. Benchmarks against the van der Pol oscillator supply external comparison, confirming the method is self-contained as a statistical inference technique without load-bearing circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities can be identified from the provided text.

pith-pipeline@v0.9.1-grok · 5655 in / 982 out tokens · 18114 ms · 2026-06-26T01:48:49.331930+00:00 · methodology

0 comments
read the original abstract

We introduce a data-driven framework that maps noisy oscillatory time series directly onto the Hopf normal form, enabling inference of underlying dynamics without knowledge of governing equations. By embedding the normal form in a probabilistic state-space model, the method jointly infers latent states and system parameters, yielding robust estimates of the natural frequency, Floquet exponent, and asymptotic phase even far from the bifurcation point and under strong noise. Combined with complex Gaussian process regression, the approach further reconstructs phase and amplitude sensitivity functions from data. Benchmarks on the van der Pol oscillator demonstrate substantially improved accuracy and noise robustness compared with existing phase-based and regression methods. This work establishes a direct bridge between normal-form theory and statistical inference, providing a general and practical route to low-dimensional descriptions of oscillatory dynamics in complex systems.

Figures

Figures reproduced from arXiv: 2606.26847 by Ryota Kobayashi, Shinsuke Koyama.

Figure 1
Figure 1. Figure 1: FIG. 1. Schematic illustration of the proposed method. (a) Noisy oscillatory time-series data observed in the measurement [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Estimation of natural frequency and Floquet exponent. Estimated natural frequency [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Denoising of oscillatory time series. The black dot [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Reconstruction of phase and amplitude sensitivity functions. Phase sensitivity function (top panels) and amplitude [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. An Isochron-Free Framework for Phase Reduction and Coupling Inference

    nlin.AO 2026-06 unverdicted novelty 6.0

    Under strong amplitude stability and near-uniform rotation, the one-period stroboscopic map of a generalized phase yields a closed circle map whose interaction term equals the coupling function of the asymptotic phase...

  2. An Isochron-Free Framework for Phase Reduction and Coupling Inference

    nlin.AO 2026-06 conditional novelty 6.0

    Under near-uniform phase rotation and strong amplitude stability, a one-period circle map of a generalized phase has the same leading-order coupling function as classical asymptotic-phase reduction, enabling isochron-...

Reference graph

Works this paper leans on

59 extracted references · 1 canonical work pages · cited by 1 Pith paper

  1. [1]

    We propose a data-driven method for construct- ing representations consistent with the Hopf nor- mal form from noisy observational time-series data, without requiring prior knowledge of the governing equations

  2. [2]

    We develop a state-space–based model for robust estimation of system parameters and latent phase– amplitude states under observation noise

  3. [3]

    The remainder of this article is organized as follows

    We infer intrinsic dynamical properties, including phase and amplitude sensitivity functions, from the inferred representations. The remainder of this article is organized as follows. Section II briefly reviews the Hopf normal form. Sec- tion III presents the proposed methodology, with detailed derivations and algorithmic procedures provided in the Append...

  4. [4]

    Our goal 4 is to computep(x t |y 1:t,Θ) fromp(x t−1|y1:t−1,Θ)

    Forward iteration (Filtering) Suppose that at timet−1 the filtering distribution is Gaussian:p(x t−1|y1:t−1,Θ) =N(µ f t−1,Σ f t−1). Our goal 4 is to computep(x t |y 1:t,Θ) fromp(x t−1|y1:t−1,Θ). We first evaluate the predictive density: p(xt |y 1:t−1,Θ) = Z dxt−1p(xt |x t−1,Θ) ×p(xt−1 |y 1:t−1,Θ).(9) Due to the nonlinearity of the state equation (5), this...

  5. [5]

    Backward iteration (Smoothing) Next, we compute the smoothing distribution p(xt|y1:T ,Θ). Starting from the final filtering distribu- tion att=T:p(x T |y1:T ,Θ), we proceed backward in time using p(xt |y 1:T ,Θ) = Z dxt+1p(xt |x t+1,y 1:t,Θ) ×p(xt+1 |y 1:T ,Θ).(12) The conditional density is given by p(xt |x t+1,y 1:t,Θ) = p(xt+1 |x t,Θ)p(x t |y 1:t,Θ) p(...

  6. [6]

    Given the esti- mated parameters ˆΘ, the latent states{µ s t }for each time series are inferred by forward–backward smoothing (see Subsection III B for details)

    Proposed method The model parameters of the state-space model Θ are first estimated from the set of time series using the EM al- gorithm (see Subsection III C for details). Given the esti- mated parameters ˆΘ, the latent states{µ s t }for each time series are inferred by forward–backward smoothing (see Subsection III B for details). Using the estimated po...

  7. [7]

    The resulting estimated limit-cycle orbit is subsequently used in (21) and (22) to estimate the PSF and ASF, respectively

    In particular, the limit-cycle orbit is estimated using (23), where the observation function is replaced with ˆh. The resulting estimated limit-cycle orbit is subsequently used in (21) and (22) to estimate the PSF and ASF, respectively. To apply CGPR for interpolating the estimated states to unobserved points in the observation space, we use the set of de...

  8. [8]

    First, a moving average with window width 0.07 is applied to reduce observation noise

    Baseline methods For comparison, we apply an empirical approach to estimate the frequency, Floquet exponent, and asymp- totic phase from noisy time-series data [30, 31]. First, a moving average with window width 0.07 is applied to reduce observation noise. The period is estimated as ˆT0 = (sn+1 −s 1)/n, wheres 1 ands n+1 denote the first and (n+ 1)-th cro...

  9. [9]

    recently proposed a method for identifying dynam- ical systems near Hopf bifurcation by analytically re- lating the phase and amplitude response curves of the normal form to those estimated directly from data. In a related direction, Arayaet al.[43] proposed a data- 9 Z1 -1 -0.5 0 0.5 1 Z2 -0.5 0 0.5 I1 -1 -0.5 0 0.5 1 I2 -1 -0.5 0 0.5 1 Z1 -1 -0.5 0 0.5 ...

  10. [10]

    To compute the mean and covariance of the filtering density (11), we approximate the joint densityp(x t,y t | y1:t−1,Θ) =p(y t |x t,Θ)p(x t |y 1:t−1,Θ) by a Gaussian

    Filtering The mean and covariance of the prediction density (9) are given by µp t =E Θ[xt |y 1:t−1] =Gµ f t−1 +E Θ[g(xt−1)|y 1:t−1],(B6) and 11 Σp t|t−1 =E Θ (xt −µ p t )(xt −µ p t )T |y 1:t−1 =E Θ (Gxt−1 +g(x t−1))(Gxt−1 +g(x t−1))T |y 1:t−1 + Σx −µ p t (µp t )T,(B7) whereE Θ[· |y 1:t−1] denotes expectation with respect to N(µ f t−1,Σ f t−1). To compute ...

  11. [11]

    Smoothing To derive the smoothing distribution (12), we approxi- mate the joint densityp(x t,x t+1 |y 1:t,Θ) by a Gaussian via moment matching: p(xt,x t+1 |y 1:t,Θ)≈ N µf t µp t+1 , Σf t Σp t,t+1 (Σp t,t+1)T Σp t+1 , (B13) where Σ p t,t+1 = cov Θ(xt,x t+1 |y 1:t). The conditional densityp(x t |x t+1,y 1:t,Θ) is then Gaussian with ˜µt =µ f t + Σp t,t+1(Σp ...

  12. [12]

    Buzsaki and A

    G. Buzsaki and A. Draguhn, Neuronal oscillations in cor- tical networks, Science304, 1926 (2004)

    1926

  13. [13]

    X. J. Wang, Neurophysiological and computational prin- ciples of cortical rhythms in cognition, Physiological Re- view90, 1195 (2010)

    2010

  14. [14]

    J. A. Mohawk and J. S. Takahashi, Cell autonomy and synchrony of suprachiasmatic nucleus circadian oscilla- tors, Trends in neurosciences34, 349 (2011)

    2011

  15. [15]

    Sch¨ afer, M

    C. Sch¨ afer, M. G. Rosenblum, J. Kurths, and H. H. Abel, Heartbeat synchronized with ventilation, Nature 392, 239 (1998)

    1998

  16. [16]

    Kralemann, M

    B. Kralemann, M. Fr¨ uhwirth, A. Pikovsky, M. Rosen- blum, T. Kenner, J. Schaefer, and M. Moser, In vivo car- diac phase response curve elucidates human respiratory heart rate variability, Nature communications4, 2418 (2013)

    2013

  17. [17]

    Kobayashi, H

    R. Kobayashi, H. Nishimaru, and H. Nishijo, Estimation of excitatory and inhibitory synaptic conductance varia- tions in motoneurons during locomotor-like rhythmic ac- tivity, Neuroscience335, 72 (2016)

    2016

  18. [18]

    J. J. Collins and I. N. Stewart, Coupled nonlinear os- cillators and the symmetries of animal gaits, Journal of Nonlinear science3, 349 (1993)

    1993

  19. [19]

    Kiehn, Locomotor circuits in the mammalian spinal cord, Annu

    O. Kiehn, Locomotor circuits in the mammalian spinal cord, Annu. Rev. Neurosci.29, 279 (2006)

    2006

  20. [20]

    Funato, Y

    T. Funato, Y. Yamamoto, S. Aoi, T. Imai, T. Aoy- agi, N. Tomita, and K. Tsuchiya, Evaluation of the phase-dependent rhythm control of human walking us- ing phase response curves, PLoS computational biology 12, e1004950 (2016)

    2016

  21. [21]

    S. H. Strogatz,Nonlinear dynamics and chaos: with ap- plications to physics, biology, chemistry, and engineering (CRC press, Boca Raton, 2018)

    2018

  22. [22]

    A. T. Winfree,The Geometry of Biological Time (Springer, New York, 2001)

    2001

  23. [23]

    Guckenheimer and P

    J. Guckenheimer and P. J. Holmes,Nonlinear oscilla- tions, dynamical systems and bifurcations of vector fields (Springer, New York, 1983)

    1983

  24. [24]

    Wiggins,Introduction to Applied Nonlinear Dynamical Systems and Chaos(Springer, New York, 2003)

    S. Wiggins,Introduction to Applied Nonlinear Dynamical Systems and Chaos(Springer, New York, 2003)

    2003

  25. [25]

    Y. A. Kuznetsov,Elements of Applied Bifurcation Theory (Springer, New York, 2004)

    2004

  26. [26]

    F. C. Hoppensteadt and E. M. Izhikevich,Weakly Con- nected Neural Networks(Springer, New York, 1997)

    1997

  27. [27]

    Wilson, Data-driven model identification near a super- critical Hopf bifurcation using phase-based approaches, Physica D476, 134635 (2025)

    D. Wilson, Data-driven model identification near a super- critical Hopf bifurcation using phase-based approaches, Physica D476, 134635 (2025)

    2025

  28. [28]

    Durbin and S

    J. Durbin and S. Koopman,Time Series Analysis by State Space Methods(Oxford University Press, Oxford, UK, 2001)

    2001

  29. [29]

    Kitagawa,Introduction to Time Series Modeling (Chapman and Hall/CRC, Boca Raton, FL , USA, 2010)

    G. Kitagawa,Introduction to Time Series Modeling (Chapman and Hall/CRC, Boca Raton, FL , USA, 2010)

    2010

  30. [30]

    Kuramoto,Chemical Oscillations, Waves and Turbu- lence(Springer, Berlin, 1984)

    Y. Kuramoto,Chemical Oscillations, Waves and Turbu- lence(Springer, Berlin, 1984)

    1984

  31. [31]

    Nakao, Phase reduction approach to synchronisation of nonlinear oscillators, Contemporary Physics57, 188 (2016)

    H. Nakao, Phase reduction approach to synchronisation of nonlinear oscillators, Contemporary Physics57, 188 (2016)

    2016

  32. [32]

    K. C. A. Wedgwood, K. K. Lin, R. Thul, and S. Coombes, Phase-amplitude descriptions of neural oscillator models, The Journal of Mathematical Neuroscience3, 2 (2013)

    2013

  33. [33]

    Shirasaka, W

    S. Shirasaka, W. Kurebayashi, and H. Nakao, Phase- amplitude reduction of transient dynamics far from at- tractors for limit-cycling systems, Chaos27, 023119 (2017)

    2017

  34. [34]

    Shirasaka, W

    S. Shirasaka, W. Kurebayashi, and H. Nakao, Phase- amplitude reduction of limit cycling systems, inThe koopman operator in systems and control: Concepts, methodologies, and applications(Springer, 2020) pp. 383– 417

    2020

  35. [35]

    M. P. Deisenroth, M. F. Huber, and U. D. Hanebeck, An- alytic moment–based gaussian process filtering, inPro- ceedings of the 26th annual international conference on machine learning(2009) pp. 225–232

    2009

  36. [36]

    Turner, M

    R. Turner, M. Deisenroth, and C. Rasmussen, State– 13 space inference and learning with Gaussian processes, in Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics(2010) pp. 868– 875

    2010

  37. [37]

    Donner, A

    C. Donner, A. Mishra, and H. Shimazaki, A projected nonlinear state–space model for forecasting time series signals, International Journal of Forecasting41, 1296 (2025)

    2025

  38. [38]

    A. P. Dempster, N. M. Laird, and D. B. Rubin, Maximum likelihood from incomplete data via the EM algorithm, Journal of the royal statistical society: series B (method- ological)39, 1 (1977)

    1977

  39. [39]

    R. H. Shumway and D. S. Stoffer, An approach to time series smoothing and forecasting using the EM algorithm, Journal of time series analysis3, 253 (1982)

    1982

  40. [40]

    Boloix-Tortosa, J

    R. Boloix-Tortosa, J. J. Murillo-Fuentes, F. J. Pay´ an- Somet, and F. P´ erez-Cruz, Complex Gaussian processes for regression, IEEE Transactions on Neural Networks and Learning Systems29, 5499 (2018)

    2018

  41. [41]

    Namura, S

    N. Namura, S. Takata, K. Yamaguchi, R. Kobayashi, and H. Nakao, Estimating asymptotic phase and amplitude functions of limit-cycle oscillators from time series data, Physical Review E106, 014204 (2022)

    2022

  42. [42]

    Yamamoto, H

    T. Yamamoto, H. Nakao, and R. Kobayashi, Gaussian process phase interpolation for estimating the asymptotic phase of a limit cycle oscillator from time series data, Chaos, Solitons & Fractals191, 115913 (2025)

    2025

  43. [43]

    A. Wolf, J. B. Swift, H. L. Swinney, and J. A. Vas- tano, Determining lyapunov exponents from a time se- ries, Physica D: Nonlinear Phenomena16, 285 (1985)

    1985

  44. [44]

    Takata, Y

    S. Takata, Y. Kato, and H. Nakao, Fast optimal en- trainment of limit-cycle oscillators by strong periodic in- puts via phase-amplitude reduction and Floquet theory, Chaos: An Interdisciplinary Journal of Nonlinear Science 31, 093124 (2021)

    2021

  45. [45]

    R. F. Gal´ an, G. B. Ermentrout, and N. N. Urban, Effi- cient estimation of phase-resetting curves in real neurons and its significance for neural-network modeling, Physi- cal Review Letters94, 158101 (2005)

    2005

  46. [46]

    Wilson, A direct method approach for data-driven in- ference of high accuracy adaptive phase-isostable reduced order models, Physica D446, 133675 (2023)

    D. Wilson, A direct method approach for data-driven in- ference of high accuracy adaptive phase-isostable reduced order models, Physica D446, 133675 (2023)

    2023

  47. [47]

    P. J. Schmid, Dynamic mode decomposition of numerical and experimental data, Journal of fluid mechanics656, 5 (2010)

    2010

  48. [48]

    I. G. K. M. O. Williams and C. W. Rowley, A data-driven approximation of the Koopman operator: Extending dy- namic mode decomposition, Journal of Nonlinear Science 25, 1307 (2015)

    2015

  49. [49]

    J. N. Kutz, S. L. Brunton, B. W. Brunton, and J. L. Proc- tor,Dynamic mode decomposition: data-driven modeling of complex systems(SIAM, 2016)

    2016

  50. [50]

    Yawata, K

    K. Yawata, K. Fukami, K. Taira, and H. Nakao, Phase autoencoder for limit-cycle oscillators, Chaos34, 063111 (2024)

    2024

  51. [51]

    Wilson, A data-driven phase and isostable reduced modeling framework for oscillatory dynamical systems, Chaos30, 013121 (2020)

    D. Wilson, A data-driven phase and isostable reduced modeling framework for oscillatory dynamical systems, Chaos30, 013121 (2020)

    2020

  52. [52]

    Wilson, Data-driven inference of high-accuracy isostable-based dynamical models in response to exter- nal inputs, Chaos31, 063137 (2021)

    D. Wilson, Data-driven inference of high-accuracy isostable-based dynamical models in response to exter- nal inputs, Chaos31, 063137 (2021)

    2021

  53. [53]

    Tanaka, S

    H.-A. Tanaka, S. Suga, A. Keida, H. Nakao, Y. Jit- sumatsu, and I. Z. Kiss, Inferring noise intensity and phase response from noisy synchronous oscillators, Phys. Rev. Res.7, 013076 (2025)

    2025

  54. [54]

    Araya, H

    Y. Araya, H. Ito, H. Kori, and H. Kitahata, Infer- ring coupled Stuart–Landau equations from waveforms, arXiv:2602.13031v1 [nlin.AO] (2026), arXiv:2602.13031 [nlin.AO]

    work page arXiv 2026

  55. [55]

    Roweis and Z

    S. Roweis and Z. Ghahramani, A unifying review of linear Gaussian models, Neural Computation11, 305 (1999)

    1999

  56. [56]

    Qui˜ nonero-Candela and C

    J. Qui˜ nonero-Candela and C. E. Rasmussen, A unifying view of sparse approximate Gaussian process regression, Journal of Machine Learning Research6, 1939 (2005)

    1939

  57. [57]

    M. Titsias, Variational learning of inducing variables in sparse Gaussian processes, inProceedings of the Twelfth International Conference on Artificial Intelligence and Statistics, Proceedings of Machine Learning Research, Vol. 5, edited by D. van Dyk and M. Welling (PMLR, Hilton Clearwater Beach Resort, Clearwater Beach, Florida USA, 2009) pp. 567–574

    2009

  58. [58]

    Solin and S

    A. Solin and S. S¨ arkk¨ a, Hilbert space methods for reduced-rank Gaussian process regression, Statistics and Computing30, 419 (2014)

    2014

  59. [59]

    S¨ arkk¨ a, A

    S. S¨ arkk¨ a, A. Solin, and J. Hartikainen, Spatiotempo- ral learning via infinite-dimensional Bayesian filtering and smoothing: A look at Gaussian process regression through Kalman filtering, IEEE Signal Processing Mag- azine30, 51 (2013)

    2013