Pith. sign in

REVIEW 3 major objections 4 minor 38 references

On Kernel Design for Regularized Volterra Series Identification of Wiener-Hammerstein Systems

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Structured kernels built from the Wiener-Hammerstein block prior identify Volterra series in cubic time and, for separable inputs, near-linear time.

desk verdict A genuinely new structural kernel design for Volterra identification, with real complexity gains, but the central PSD theorem has a fixable gap on negative lags. read the letter →

arxiv 2505.20747 v1 pith:ZESQ2NWU submitted 2025-05-27 eess.SY cs.SY

Yu Xu , Biqiang Mu , Tianshi Chen This is my paper · ORCID
classification eess.SYcs.SY MSC 93B3093C10
keywords Wiener-HammersteinsystemsVolterraseriesidentificationkernel-basedregularizationkerneldesignoutputmatrixseparabilityrank
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that the bottleneck of regularized Volterra-series identification—kernel design and computation—can be resolved by building the kernel from the actual Wiener-Hammerstein (WH) block structure instead of from generic smoothness assumptions. For a WH system with finite-memory linear blocks and polynomial nonlinearity, the true Volterra maps factor as $h_m(t_1,\ldots,t_m)=a_m\sum_{\tau=0}^{n-1}g_2(\tau)\prod_{i=1}^m g_1(t_i-\tau)$, and the paper turns this factorization into a positive-semidefinite kernel with nonzero off-diagonal blocks coupling different Volterra orders. The convolution structure of the kernel makes the output covariance matrix computable in $O(N^3)$ time, and in a Wiener-system special case with separable input signals the cost drops to $O(N\gamma^2)$, where $\gamma$ is an explicit rank. Monte Carlo simulations show better average prediction accuracy than generic smoothness kernels and than Bayesian semiparametric and prediction-error baselines on the tested systems.

What carries the argument

The load-bearing object is the factored Volterra-map identity $h_m(t_1,\ldots,t_m)=a_m\sum_{\tau=0}^{n-1}g_2(\tau)\prod_{i=1}^m g_1(t_i-\tau)$, which expresses every order of a Wiener-Hammerstein system as a shared convolution with $g_2$ of products of $g_1$. The designed kernel (10a)-(10b) replaces $g_1,g_2$ by kernels $\kappa_1,\kappa_2$ and $g_1$ by $\zeta$; Theorem 3.1 guarantees the full kernel matrix is positive semidefinite exactly when $\kappa_1(t,s)-\zeta(t)\zeta(s)$ is positive semidefinite. Computation is carried by two structural identities: $Q=\mathrm{conv2}(K_2,Q_w)$ for the output kernel, and, in the Wiener special case, the separability decomposition $Q=\bar{U}\bar{V}^T$ whose rank $\gamma$ is given by the combinatorial formula in Theorem 4.2.

What would settle it

For the Wiener special case $\kappa_2(t,s)=1$ when $t=s=0$ and $0$ otherwise, take input $u(t)=e^{\lambda t}$ (so $r=1$) and $M=3$; Theorem 4.2 says the output kernel matrix $Q_w$ has exact separability rank $\gamma=5$. Compute the singular values of the numerically formed $Q_w$ for $N$ large: if the sixth singular value is not at machine precision relative to the largest, the rank formula fails. The same check applies for $u(t)=e^{-0.0003t}\cos(0.1t+\pi/3)$ with $r=2$ and $M=3$, where the predicted rank is $\gamma=14$.

Watch

Extended reading notes

Core claim

The central claim is that the optimal-kernel identity $P_{\mathrm{opt}}=\theta_0\theta_0^T$, applied to the exact Volterra-map form of a WH system, produces an explicit cross-kernel $$K_{pq}(t_p,s_q;\eta_{pq})=\sum_{\xi_1=0}^{n-1}\sum_{\xi_2=0}^{n-1}\kappa_2(\xi_1,\xi_2;\eta_2)K^w_{pq}(t_p-\xi_1,s_q-\xi_2;a_p,a_q,\eta_1),$$ where $\kappa_1,\kappa_2$ are user-designed kernels for the two linear blocks and $\zeta$ is a designed stand-in for $g_1$. This kernel has nonzero off-diagonal blocks, so it encodes correlations between different Volterra orders, and it is positive semidefinite whenever $\kappa_1(t,s)-\zeta(t)\zeta(s)$ is positive semidefinite. The output kernel matrix has the convolution form $Q=\mathrm{conv2}(K_2,Q_w)$, computable in $O(N^3)$; in the special case $\kappa_2(t,s)=1$ if $t=s=0$ and $0$ otherwise, with inputs satisfying $u(t-b)=\sum_{i=1}^r\pi_i(t)\rho_i(b)$, $Q$ is exactly separable with rank $\gamma=\frac{(r+M-1)!}{(r-1)!M!}+2\sum_{m=1}^{M-1}\frac{(r+m-1)!}{(r-1)!m!}$, reducing the empirical-Bayes and prediction computations to $O(N\gamma^2)$.

Load-bearing premise

The load-bearing premise is that the true system is exactly a Wiener-Hammerstein system with finite-memory linear blocks of length $n$ and a static nonlinearity that is a polynomial of degree $M$; if the real dynamics have infinite memory or a non-polynomial nonlinearity, the kernel encodes the wrong structure and can introduce bias.

Editorial extensions

If this is right

  • For Wiener-Hammerstein systems with polynomial nonlinearity, the regularization kernel can encode block-structure and cross-order correlations rather than only smoothness and decay.
  • The empirical-Bayes estimate and hyperparameter criterion can be evaluated in $O(N^3)$ time independent of the number of Volterra coefficients $n_\theta$.
  • For Wiener systems with separable input signals, the output kernel matrix has explicit rank $\gamma$ and the computation scales as $O(N\gamma^2)$, i.e., linearly in sample size for fixed $\gamma$.
  • On the tested Wiener-Hammerstein and Wiener databanks, the proposed kernels give better average prediction fit than the SED-MPK kernel, and better robustness than Bayesian semiparametric and prediction-error baselines in the D3 comparison.
  • Hyperparameters are still tuned by empirical Bayes, preserving automatic model-complexity control in short or low-SNR data.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same kernel construction should extend to Hammerstein systems and to parallel interconnections of Wiener-Hammerstein blocks, because the Volterra map of a parallel structure is a sum of terms of type (8) and the kernel would be the corresponding sum, keeping the convolution form.
  • Because $\gamma$ depends only on $M$ and $r$, the $O(N\gamma^2)$ regime is best for weakly nonlinear systems or low-complexity inputs; a testable extension is to update the separable generators incrementally for online or adaptive identification.
  • When the FIR/polynomial prior is wrong, a natural safeguard is to add a generic smoothness kernel term to the designed kernel, giving an estimator that can fall back to smoothness when the block-structure prior is misspecified.
  • The comparisons emphasize prediction and block fits; a further useful test is whether the off-diagonal blocks recover the true cross-correlations between Volterra orders, something the block-diagonal baselines cannot do.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper addresses kernel-based regularized identification of Volterra series for Wiener-Hammerstein (WH) systems with polynomial nonlinearity. Starting from the optimal kernel θ_p θ_q^T, the authors design structural kernels K_pq of the form (10a)-(10b), which embed the WH convolution structure through κ2 and products of κ1 and ζ. Theorem 3.1 states a positive semidefiniteness (PSD) condition for the resulting kernel matrix; Theorem 4.1 gives an O(N^3) construction of the output kernel matrix Q via a 2-D convolution; Theorem 4.2 shows that, for a delta-like κ2 and inputs satisfying a separability condition (15), Q has separability rank γ and the cost drops to O(Nγ^2). The paper includes Monte Carlo experiments on four databanks, comparing the proposed kernels against SED-MPK and two existing Wiener-system methods.

Significance. If the proof gaps identified below are fixed, this is a valuable contribution: it provides a direct, structural kernel for Volterra maps of WH systems with nonzero off-diagonal blocks, and it matches the state-of-the-art O(N^3) complexity while encoding more prior knowledge than previous kernels. The separability-rank analysis in Theorem 4.2 is nontrivial and the simulations support the practical usefulness of the approach. The authors provide detailed appendix proofs and a transparent experimental setup, which are strengths. The main weaknesses are the incomplete PSD guarantee for the implemented kernel at negative lags and the underspecified convolution in Theorem 4.1; both appear fixable, which is why I recommend major revision rather than rejection.

major comments (3)
  1. [Section 3, Eq. (10a), and Theorem 3.1] The proposed kernel evaluates κ1(t_i−ξ1, s_i−ξ2) and ζ at arguments that can be negative whenever ξ1>t_i or ξ2>s_i, since ξ1, ξ2 range over 0,...,n−1 while t_i, s_i start at 0. However, κ1 and ζ are introduced in Section 2.1 only on R_+, and the proof of Theorem 3.1 in Appendix A.2 establishes PSD of P^w only for nonnegative indices. If the DC formulas are naively extended to negative arguments, the sufficient condition of Theorem 3.1 fails: for ζ as in Corollary 3.1(1), κ1(−r,−r)−ζ(−r)^2 = c1^2 e^{2αr}(1−e^{2βr}) < 0 for r>0. Thus the PSD guarantee does not cover the implemented kernel whenever κ2 has nonzero off-diagonal entries, which is exactly the D1/D2 setting. The authors should specify a causal zero extension (or another explicit extension) for κ1, ζ, and Kw at negative lags and then prove that the convolution in (10a) preserves PSD under that extension; as written, the claim that P defined through (10) is PSD is not established.
  2. [Section 4.1, Eq. (11), and Theorem 4.1] With K2 ∈ R^{n×n} and Qw ∈ R^{N×N}, the 2-dimensional convolution conv2(K2, Qw) as defined in the footnote has size (N+n−1)×(N+n−1), not N×N as stated for Q. Moreover, Eqs. (30)-(32) require Qw_pq(t−ξ1, s−ξ2) at negative indices when t or s is near 1, but Qw in (12) is only defined for indices 1,...,N. The theorem should specify the zero-padding convention (for example, u(t)=0 for t≤0, corresponding to the pre-windowing option in footnote 2) and state which submatrix of the convolution is taken. Without this, the claimed O(N^3) construction of Q is not well defined.
  3. [Section 5, databanks D1-D3] The structural prior (8) assumes FIR linear blocks G1 and G2 and a polynomial nonlinearity, but Databanks D1 and D2 use IIR linear blocks and Databank D3 uses a saturation nonlinearity. The only justification is the qualitative truncation and polynomial-approximation argument in Remarks 2.1 and 2.2. Since these experiments are used as evidence for the kernel's effectiveness, the paper should either provide a quantitative statement about the resulting approximation error or explicitly present these cases as robustness checks of a misspecified model rather than as instances where the structural prior holds.
minor comments (4)
  1. [Abstract and Theorem 4.2] The abstract states that the complexity is lowered to O(Nγ^2), but Theorem 4.2 gives O(Nγ^2+n^2r) and requires the additional assumption that κ1 is extended-p semiseparable in order to remove the n^2r term; the abstract should state this extra condition.
  2. [Corollary 3.1(2) and Appendix A.3] The eigenfunction notation is inconsistent: the corollary statement uses ψ_i while the proof uses ϕ_i for the same objects, and the claim l≪N is not satisfied in the simulations of Databank D2, where l=100 and N=400.
  3. [Section 5.1] The label 'R VS' is used without definition; it should be spelled out on first use, for example 'regularized Volterra series (R VS)'.
  4. [Section 4.2, Eq. (15)] It would be useful to state explicitly which test inputs satisfy the separability assumption; white Gaussian noise (D2) does not satisfy it exactly, while the damped cosine of D4 does with small r.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: kernel (10) is built from the WH structural prior (8) via the standard optimal-kernel plug-in guideline; hyper-parameters are tuned on training data, all fits are computed on held-out test data (Sec. 5.1-5.3); same-author citations [17,33] are non-load-bearing building blocks (the key bound (29) is proven in-paper); no fitted value is relabeled as a prediction.

full rationale

The central claims — kernel (10) encodes the WH structure (8), Theorem 3.1's PSD condition, Theorem 4.1's O(N^3) construction, and Theorem 4.2's separability rank γ with O(Nγ^2) complexity — are derived in the paper rather than imported. Kernel (10) is obtained from Proposition 3.1 by the standard plug-in procedure (replace g1^0, g2^0 by κ1, κ2, ζ), i.e., it is constructed from the model class, not from the data used to evaluate it: Section 5.1 explicitly splits the noiseless data into training and test sets, the EB cost (5) is minimized on training data, and PFit/GFit/NFit (19)-(20) are computed on held-out test data, so no fitted hyper-parameter is relabeled as a prediction. The same-author citations ([17] for Lemma 2.1 and the DC kernel, [28] for the design guideline, [33] for the DC-kernel Mercer expansion, [34]/[38] for semiseparable complexity) do not carry the conclusions: Lemma 2.1 is used only as motivation ('Clearly, P^opt cannot be applied in practice'), and Corollary 3.1(2) proves in-paper the bound ∥λ∥² = 8/π²·Σ_i 1/(2i−1)² < 1 (Eq. 29) on top of the parameter-free eigenfunction expansion of [33]. Remark 4.2 explicitly concedes that the diagonal special case of (10) has an output kernel equivalent to the known SED-MPK [9, Eq. (36)], so no known result is relabeled. The skeptic's negative-lag concern — the A.2 proof covers K^w at nonnegative lags while the convolution in (10a) evaluates κ1, ζ at negative lags, where the DC extension gives κ1(−r,−r)−ζ(−r)² < 0 — is a possible correctness gap in Theorem 3.1, not circularity: it involves no fitted parameter, no self-citation chain, and no definitional equivalence, and it does not affect Theorem 4.1's input-shifting construction or Theorem 4.2's rank algebra. Score 2 reflects several same-author building-block citations that are instrumental but non-load-bearing; the derivation itself is self-contained and benchmarked against external methods ([9,14,22]).

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

The paper builds on standard kernel and regularization results and on the WH representation (8). The main assumptions are the WH/polynomial/FIR structure, the DC kernel as a prior for stable overdamped linear blocks, and the separability condition (15) for the fast algorithm. The kernel hyperparameters are not fixed by theory and must be tuned, which is the intended use of empirical Bayes.

free parameters (6)
  • DC kernel parameters (c1, α1, β1) for κ1
    Shape the prior for the first linear block in (10b); estimated by empirical Bayes (5), not fixed by the theory.
  • DC kernel parameters (c2, α2, β2) for κ2
    Shape the prior for the second linear block in (10a); estimated by empirical Bayes.
  • Polynomial coefficients a_m, m=0,...,M
    Appear as scaling hyperparameters in (10b) and Remark 3.2; estimated by empirical Bayes along with other hyperparameters.
  • Noise variance σ²
    A free parameter estimated jointly in the empirical Bayes cost (5).
  • Number of basis functions l for ζ in Corollary 3.1 case 2 = 100 in D2 and D4 simulations
    User-chosen truncation; larger l gives a richer off-diagonal prior at extra computational cost.
  • Separability rank r of the input representation (15) = 2 for the D4 input u(t)=e^{-0.0003t} cos(0.1t+π/3); otherwise assumed small
    Determines γ in Theorem 4.2; not estimated from data, but assumed by the user for the fast algorithm.
assumptions (6)
  • domain assumption The true system is exactly a Wiener-Hammerstein system with a polynomial static nonlinearity (7) and FIR linear blocks of order n, giving Volterra maps (8).
    This is the starting assumption in Section 2.2 and underlies kernel (10); if the linear blocks are IIR or the nonlinearity is not polynomial, the kernel is misspecified.
  • domain assumption The finite Volterra series (1) is a good truncation of the infinite Volterra series, with bias made small by choosing n and M large.
    Used in Remark 2.1 to justify applying the FIR-derived kernel to systems with longer memory; the bias is not quantified.
  • domain assumption The linear blocks are stable and overdamped so that the DC kernel (6) is an appropriate prior.
    Used in Corollary 3.1 to select κ1 and κ2 as DC kernels; for resonant or unstable blocks the proposed ζ choices may not be appropriate.
  • domain assumption The input signal satisfies (15) for Theorem 4.2, and the input is persistently exciting.
    The separability rank and O(Nγ²) result depend on this input structure; not all common inputs satisfy it with small r.
  • domain assumption Measurement noise v(t) is i.i.d. Gaussian with zero mean and variance σ².
    Stated in Section 2; the estimator and empirical Bayes cost (5) are derived under this noise model.
  • standard math Standard matrix identities and kernel theory: Schur product theorem, convolution theorem, multinomial theorem, matrix inversion lemma, determinant lemma, and the eigendecomposition of the DC kernel from [33].
    Used in the proofs of Theorems 3.1, 4.1, and 4.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On Kernel Design for Regularized Volterra Series Identification of Wiener-Hammerstein Systems." pith.science (2026). https://pith.science/paper/ZESQ2NWU

@misc{pith2026250520747,
  author       = {Pith},
  title        = {Pith review of: On Kernel Design for Regularized Volterra Series Identification of Wiener-Hammerstein Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZESQ2NWU}},
  note         = {Machine review of arXiv:2505.20747}
}
abstract

There have been increasing interests on the Volterra series identification with the kernel-based regularization method. The major difficulties are on the kernel design and efficiency of the corresponding implementation. In this paper, we first assume that the underlying system to be identified is the Wiener-Hammerstein (WH) system with polynomial nonlinearity. We then show how to design kernels with nonzero off-diagonal blocks for Volterra maps by taking into account the prior knowledge of the linear blocks and the structure of WH systems. Moreover, exploring the structure of the designed kernels leads to the same computational complexity as the state-of-the-art result, i.e., $O(N^3)$, where $N$ is the sample size, but with a significant difference that the proposed kernels are designed in a direct and flexible way. In addition, for a special case of the kernel and a class of widely used input signals, further exploring the separable structure of the output kernel matrix can lower the computational complexity from $O(N^3)$ to $O(N\gamma^2)$, where $\gamma$ is the separability rank of the output kernel matrix and can be much smaller than $N$. We finally run Monte Carlo simulations to demonstrate the proposed kernels and the obtained theoretical results.

Figures

Figures reproduced from arXiv: 2505.20747 by the authors.

Figure 1
Figure 1. The block diagram of the WH system where [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 1
Figure 1. Accordingly, the output kernel matrix Q in (11) becomes Qw in (12), i.e., Q = Qw. In this case, we show below that Q can have the separable structure, which can be used to calculate (4b) and (5) in a more efficient way. Before proceeding to the details, we make the following assumption on the input signal u(t) that there exist πi , ρi : R+ → R, i = 1, . . . , r with r ∈ N such that u(t − b) = Xr i=1 πi(t)ρi(b). (15)… view at source ↗
Figure 2
Figure 2. The block-oriented nonlinear system in D1. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figures from the paper (4 more)
Figure 3
Figure 3. Figure 3: Boxplot of the prediction fits of the estimators based [PITH_FULL_IMAGE:figures/full_fig_p009_3.png]
Figure 4
Figure 4. Figure 4: Boxplot of the prediction fits of the estimators based [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Boxplot of the prediction fits of the estimators based [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 7
Figure 7. Figure 7: Average computation time (in seconds) of evaluating [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    Boyd and L

    S. Boyd and L. Chua. Fading memory and the problem of approximating nonlinear operators with Volterra series. IEEE Transactions on Circuits and Systems , 32(11):1150– 1161, 1985

    work page 1985

  2. [2]

    Pillonetto, F

    G. Pillonetto, F. Dinuzzo, T. Chen, G. De Nicolao, and L. Ljung. Kernel methods in system identification, machine learning and function estimation: A survey. Automatica, 50(3):657–682, 2014

    work page 2014

  3. [3]

    A. Chiuso. Regularization and Bayesian learning in dynamical systems: Past, present and future.Annual Reviews in Control, 41:24–38, 2016

    work page 2016

  4. [4]

    Ljung, T

    L. Ljung, T. Chen, and B. Mu. A shift in paradigm for system identification. International Journal of Control , 93(2):173– 180, 2020

    work page 2020

  5. [5]

    Pillonetto, T

    G. Pillonetto, T. Chen, A. Chiuso, G. De Nicolao, and L. Ljung. Regularized System Identification: Learning Dynamic Models from Data . Springer Nature, 2022

    work page 2022

  6. [6]

    J. G. Stoddard, J. S. Welsh, and H. Hjalmarsson. EM-based hyperparameter optimization for regularized Volterra kernel estimation. IEEE Control Systems Letters , 1(2):388–393, 2017

    work page 2017

  7. [7]

    Birpoutsoukis, A

    G. Birpoutsoukis, A. Marconato, J. Lataire, and J. Schoukens. Regularized nonparametric Volterra kernel estimation. Automatica, 82:324–327, 2017

    work page 2017

  8. [8]

    Birpoutsoukis, P

    G. Birpoutsoukis, P. Z. Csurcsia, and J.Schoukens. Efficient multidimensional regularization for Volterra series estimation. Mechanical Systems and Signal Processing , 104:896–914, 2018

    work page 2018

Show all 38 references
  1. [9]

    Dalla Libera, R

    A. Dalla Libera, R. Carli, and G. Pillonetto. Kernel- based methods for Volterra series identification. Automatica, 129:109686, 2021

    2021

  2. [10]

    G. Palm. On representation and approximation of nonlinear systems. part ii: Discrete time. Biol. Cybern., 34(1):49 – 52, 1979

    1979

  3. [11]

    Bai and F

    E.W. Bai and F. Giri. Block-oriented Nonlinear System Identification. Springer London, London, 2010

    2010

  4. [12]

    Schoukens and L

    J. Schoukens and L. Ljung. Nonlinear system identification: A user-oriented road map. IEEE Control Systems Magazine, 39(6):28–99, 2019

    2019

  5. [13]

    Enqvist and L

    M. Enqvist and L. Ljung. Linear approximations of nonlinear FIR systems for separable input processes. Automatica, 41(3):459–473, 2005. Data-Based Modelling and System Identification

    2005

  6. [14]

    Hagenblad, L

    A. Hagenblad, L. Ljung, and A. Wills. Maximum likelihood identification of Wiener models. Automatica, 44(11):2697– 2705, 2008

    2008

  7. [15]

    Tiels and J

    K. Tiels and J. Schoukens. Wiener system identification with generalized orthonormal basis functions. Automatica, 50(12):3147–3154, 2014

    2014

  8. [16]

    Schoukens and K

    M. Schoukens and K. Tiels. Identification of block-oriented nonlinear systems starting from linear approximations: A survey. Automatica, 85:272–292, 2017

    2017

  9. [17]

    T. Chen, H. Ohlsson, and L. Ljung. On the estimation of transfer functions, regularizations and Gaussian processes- revisited. Automatica, 48(8):1525–1535, 2012

    2012

  10. [18]

    Pillonetto and A

    G. Pillonetto and A. Chiuso. Tuning complexity in regularized kernel-based regression and linear system identification: The robustness of the marginal likelihood estimator. Automatica, 58:106–117, 2015. 16

    2015

  11. [19]

    B. Mu, T. Chen, and L. Ljung. On asymptotic properties of hyperparameter estimators for kernel-based regularization methods. Automatica, 94:381–395, 2018

    2018

  12. [20]

    Y. Ju, B. Mu, L. Ljung, and T. Chen. Asymptotic theory for regularized system identification part I: Empirical Bayes hyperparameter estimator. IEEE Transactions on Automatic Control, 68(12):7224–7239, 2023

    2023

  13. [21]

    Pillonetto and A

    G. Pillonetto and A. Chiuso. Gaussian processes for Wiener- Hammerstein system identification. IF AC Proceedings Volumes, 42(10):838–843, 2009. 15th IF AC Symposium on System Identification

    2009

  14. [22]

    Lindsten, T

    F. Lindsten, T. B. Sch¨ on, and M. I. Jordan. Bayesian semiparametric Wiener system identification. Automatica, 49(7):2053–2063, 2013

    2013

  15. [23]

    R. S. Risuleo, G. Bottegal, and H. Hjalmarsson. A nonparametric kernel-based approach to Hammerstein system identification. Automatica, 85:234–247, 2017

    2017

  16. [24]

    R. S. Risuleo, F. Lindsten, and H. Hjalmarsson. Bayesian nonparametric identification of Wiener systems. Automatica, 108:108480, 2019

    2019

  17. [25]

    N. Wiener. Nonlinear problems in random theory . Wiley, New York, 1958

    1958

  18. [26]

    Schetzen

    M. Schetzen. The Volterra and Wiener Theories of Nonlinear Systems. John Wiley & Sons, 1980

    1980

  19. [27]

    L. Ljung. System identification: theory for the user . Upper Saddle River, NJ: Prentice Hall, 12 1999

    1999

  20. [28]

    T. Chen. On kernel design for regularized LTI system identification. Automatica, 90:109–122, 2018

    2018

  21. [29]

    Schoukens, R

    J. Schoukens, R. Pintelon, and M. Enqvist. Study of the LTI relations between the outputs of two coupled Wiener systems and its application to the generation of initial estimates for Wiener–Hammerstein systems. Automatica, 44(7):1654– 1665, 2008

    2008

  22. [30]

    Nowak and B.D

    R.D. Nowak and B.D. Van Veen. Random and pseudorandom inputs for Volterra filter identification. IEEE Transactions on Signal Processing, 42(8):2124–2135, 1994

    1994

  23. [31]

    Birpoutsoukis

    G. Birpoutsoukis. Volterra series estimation in the presence of prior knowledge . PhD thesis, Vrije Universiteit Brussel Brussel, Belgium, 2018

    2018

  24. [32]

    W. K. Wong. Comparing robust properties of A, D, E and G- optimal designs. Computational Statistics & Data Analysis , 18(4):441–448, 1994

    1994

  25. [33]

    T. Chen. Continuous-time DC kernel—a stable generalized first-order spline kernel. IEEE Transactions on Automatic Control, 63(12):4442–4447, 2018

    2018

  26. [34]

    Chen and M

    T. Chen and M. S. Andersen. On semiseparable kernels and efficient implementation for regularized system identification and function estimation. Automatica, 132:109682, 2021

    2021

  27. [35]

    M. Zorzi. Nonparametric identification of Kronecker networks. Automatica, 145:110518, 2022

    2022

  28. [36]

    R. J.G.B. Campello, G. Favier, and W. C. do Amaral. Optimal expansions of discrete-time Volterra models using Laguerre functions. Automatica, 40(5):815–822, 2004

    2004

  29. [37]

    Kekatos and G

    V. Kekatos and G. B. Giannakis. Sparse Volterra and polynomial regression models: Recoverability and estimation. IEEE Transactions on Signal Processing , 59(12):5907–5920, 2011

    2011

  30. [38]

    M. S. Andersen and T. Chen. Smoothing splines and rank structured matrices: Revisiting the spline kernel. SIAM Journal on Matrix Analysis and Applications, 41(2):389–412, 2020. 17

    2020

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.