Tensor Network Kernel Machines: A JAX Framework for Machine Learning and Nonlinear System Identification

Albert Saiapin; Kim Batselier

REVIEW 3 major objections 4 minor 50 references

Tensor Network Kernel Machines: A JAX Framework for Machine Learning and Nonlinear System Identification

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read An open-source Python library makes tensor network kernel machines competitive with established nonlinear system identification baselines while training in under a second to a few seconds.

desk verdict A solid, well-scoped software paper for a JAX TNKM library; the empirical claims are plausible but the benchmark comparisons are only as good as the external baselines they inherit. read the letter →

arxiv 2608.07043 v1 pith:ZVC6D7B2 submitted 2026-08-07 cs.MS cs.LGcs.SYeess.SY

Albert Saiapin , Kim Batselier This is my paper · ORCID

classification cs.MScs.LGcs.SYeess.SY

keywords tensornetworkkernelmachinesnonlinearsystemidentificationJAXtensor-traindecompositioncanonicalpolyadicalternatingleastsquaresNARXmodelsPythonlibrary

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

Tensor network kernel machines predict a target by combining per-input nonlinear features through a tensor product and storing the resulting weight vector as a low-rank tensor network. This paper introduces tnkm, an open-source Python library that unifies the components of this model family—polynomial, Fourier, and Volterra feature maps, CP and tensor-train parameterizations, and alternating least squares or gradient-based training—under one interface. The paper's central claim is that these implemented models achieve predictive accuracy comparable to established black-box and gray-box methods on the Coupled Electric Drives, Cascaded Tanks, and Silverbox benchmarks, while training in under a second to a few seconds on a laptop. If correct, this gives practitioners a reproducible and customizable starting point for tensor-network-based nonlinear modeling and system identification.

What carries the argument

The load-bearing object is the TNKM itself: a model $f(x)=\phi(x)^\top w$ in which $\phi$ is a tensor product of one-dimensional feature maps and $w$ is reinterpreted as a $D$-way tensor stored in either CP or tensor-train form. This storage is what converts an exponential parameter count into one that grows linearly in $D$, and the same multilinear structure is what alternating least squares exploits by solving one core at a time. On the benchmarks the model is trained as a NARX system with min-max normalized inputs and evaluated by recursive free-run simulation, so the reported RMSEs are over multi-step predicted trajectories rather than one-step-ahead errors.

What would settle it

Run the released code on the three benchmarks while recomputing all baseline RMSE values under the exact min-max normalization, lag sets, and recursive free-run simulation protocol stated in Section 4.2; if any baseline moves enough to push TNKM outside the reported competitive band, the central claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that a unified software implementation makes tensor network kernel machines practical and competitive for nonlinear regression and system identification. A TNKM is the linear-in-features model $f(x)=\phi(x)^\top w$ with $\phi(x)=\phi^{(D)}(x_D)\otimes\cdots\otimes\phi^{(1)}(x_1)$, and $w$ stored as a CP or tensor-train decomposition, so a parameter vector that would otherwise grow exponentially is compressed to one that grows linearly in the number of inputs. On the Coupled Electric Drives benchmark the CP model with rank 15 and polynomial degree-3 features reaches RMSE 0.085/0.077 in under 0.1 s; on Cascaded Tanks the same backbone with rank 13 and degree-2 features reaches RMSE 0.347 in under 0.1 s; on Silverbox the tensor-train model with rank 4 and degree-2 features reaches $\mathrm{RMSE}\times 10^3$ values of 0.402/0.881/0.368 in one to sixty seconds. These numbers sit close to GPNARX and SUBNET and slightly behind dynoNet and physics-informed NL-LFR, while training orders of magnitude faster. The paper also shows ALS converging close to a kernel ridge regression reference on Airfoil Self-Noise in roughly a third of the time Adam needs.

Load-bearing premise

The entire competitive-accuracy claim depends on the baseline RMSE and training-time values taken from the official Nonlinear Benchmark website being directly comparable to the TNKM results, under identical data splits, preprocessing, and evaluation protocol.

Editorial extensions

If this is right

On the three benchmarks, TNKM reaches RMSE comparable to GPNARX and SUBNET while training on a laptop in less than a second to a few seconds, making broad hyperparameter search affordable.
Because the interface separates features, tensor backbones, and optimizers, a practitioner can switch between polynomial, Fourier, and Volterra features or between CP and tensor-train models without rewriting the training loop.
ALS offers a fast deterministic default for least-squares TNKM training, approaching the kernel ridge regression reference on Airfoil Self-Noise in about one third of the time Adam requires; gradient-based methods remain available for other losses.
The tensor-train model on Silverbox maintains stable free-run predictions over roughly 35,000 simulation steps, indicating the compact rank-4 parameterization does not accumulate excessive simulation error.
Releasing the code and experiments means the reported tables can be reproduced and extended by independent users.

Reading between the lines

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

The competitive numbers depend on the official benchmark website's baselines using the same normalization, lag sets, and recursive-simulation protocol; if they do not, the comparison would need to be redone under a shared protocol.
Since feature maps are predefined rather than learned, TNKM accuracy likely transfers best to systems whose nonlinearities resemble polynomial, Fourier, or Volterra structure; a stress test on discontinuous or highly oscillatory dynamics would clarify the boundary.
The reported training times come from one laptop CPU, so scaling to GPUs, very large datasets, or high tensor ranks is not established by this paper.
The modular design invites testing additional backbones such as tensor ring, Tucker, and hierarchical Tucker, which the paper lists as future work rather than demonstrated capabilities.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

This is a software contribution, not a new method. The TNKM equations are the authors' prior work, and the paper says so openly. What's new is the unified JAX implementation, the modular API, and the benchmark evaluation. On the merits, that's a useful and honest contribution.

The library design is clean: feature maps, CP/TT backbones, and ALS and gradient optimizers are separated behind a common interface. The Airfoil optimization study is a nice sanity check—ALS beats Adam on time and matches KRR. The system identification results on Coupled Electric Drives, Cascaded Tanks, and Silverbox are competitive with published baselines, and the training times are genuinely impressive (<0.1 sec to a few seconds).

The main weakness is benchmark comparability. TNKM numbers come from the authors' own protocol—min-max normalization on training data, heuristic NARX lags, free-run simulation seeded with 10 or 50 measured outputs—while the baseline numbers are copied from the Nonlinear Benchmark website. The paper doesn't demonstrate that the website baselines used the same normalization, the same simulation initialization, or the same evaluation mode. If the protocols differ, the RMSE comparisons in Tables 2–4 are not apples-to-apples. This is a load-bearing assumption, and I think the stress-test note is right to flag it. Also, all results are single runs with no error bars; for a library paper that claims robustness, that's a minor but real gap. The hyperparameters are selected on validation data, which is fine, but the lag sets are heuristic—fine for a demo, less fine for a strong claim.

I don't think the circularity concern holds: they restate their own prior formulation, but that's disclosed and expected for a software paper. The citation pattern is fine. The code is promised under MIT with an experiments directory; no commit hash, but that's a minor reproducibility nit.

The paper is for people who want to use or extend TNKM without reimplementing it from scratch. It's a fair software paper, worth refereeing. I'd like to see the authors either re-run the baselines under their protocol or show that the website protocol matches, and I'd want error bars or at least multiple seeds. Send it to peer review; it deserves a serious referee who will check the benchmark protocol.

Referee Report

3 major / 4 minor

Summary. The manuscript presents tnkm, an open-source JAX-based Python library for tensor-network kernel machines (TNKMs), with modular feature maps, CP/TT parameterizations, and ALS/gradient optimizers. It evaluates the library on an Airfoil regression task comparing ALS and Adam against KRR, and on three nonlinear system identification benchmarks (Coupled Electric Drives, Cascaded Tanks, Silverbox) using NARX models with recursive simulation. The paper claims competitive accuracy with compact parameters and training times on the order of seconds or less, based on RMSE comparisons with baseline results taken from the official Nonlinear Benchmark website.

Significance. If the reported accuracy and speed hold, tnkm provides a practical, extensible framework that complements existing tensor-network software. The code is publicly available under a permissive license, and the modular design is a genuine contribution. The main load-bearing issue is the protocol comparability of the baseline numbers; without resolving it, the empirical claims are not yet fully supported.

major comments (3)

[Section 4.2, Tables 2-4] The central claim that TNKM is 'comparable to the best-performing methods' is based on RMSEs copied from the official Nonlinear Benchmark website, while the TNKM results are produced in-house with min-max normalization, heuristic NARX lags, and recursive simulation initialized with the first 10 or 50 measured output samples. The manuscript provides no evidence that the website baseline RMSEs were computed under the same normalization, train/test split, initialization length, or recursive simulation mode; if any baseline used one-step-ahead prediction or a different output scaling, the RMSE values would not be commensurate. Please report the exact website evaluation protocol and verify that it matches the TNKM protocol, or re-run the baselines under the same protocol; otherwise the accuracy claims should be restricted to internal TNKM comparisons.
[Section 4.2, Tables 2-4] The training-time comparisons are not apples-to-apples: TNKM times are measured on a Dell Latitude laptop, while baseline times are reported as broad ranges ('1-60 sec', '1-60 min') taken from the Nonlinear Benchmark website, with no statement about the hardware, software, or implementation used for those baselines. Thus the headline 'less than one second' versus '1-60 sec' may reflect hardware differences rather than algorithmic efficiency. Please report measured training times for all methods on the same hardware, or explicitly qualify the comparison as indicative only.
[Sections 4.1 and 4.2] All experimental results consist of single runs. In Table 1, the ALS and Adam comparisons use fixed seeds, and Tables 2-4 report one RMSE value per test set, with no variance across random initializations, data resampling, or hyperparameter perturbations. It is therefore unclear whether observed differences (for example, TNKM 0.347 versus CT-SUBNET 0.306 in Table 3) are meaningful. Please provide means and standard deviations over at least several independent runs, and for the Airfoil experiment over multiple random train/validation splits.

minor comments (4)

[Section 4.1, Table 1] ALS achieves a validation MSE of 0.102, which is lower than the KRR baseline of 0.15, even though the text describes KRR as the unconstrained solution of the underlying regression problem; please clarify the setting (for example, different regularization, feature dimension, or data split) that makes this comparison meaningful.
[Section 4.2] Hyperparameter selection is described only as 'heuristically based on partial autocorrelation' and 'cross-validation on the training (validation) data'; please report the cross-validation folds, search grid, criterion, and the exact chosen values for ranks, feature degrees, lag sets, and regularization parameters for each benchmark so that the results are reproducible.
[Section 4.1] There is a typo in the configuration description: 'R=ℓ 2 regularization of the tensor cores v' should read 'with ℓ2 regularization of the tensor cores v'.
[Table 4] The TNKM training time is listed as '1-60 sec', unlike '<0.1 sec' in Tables 2 and 3, while the text says 'only several seconds'; please clarify the actual measured training time and why it differs from the other benchmarks.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation chain found: the TNKM results are validated against external benchmark data and no fitted quantity is renamed as a prediction.

full rationale

The paper is a software-framework contribution rather than a derivation of a new mathematical result. The TNKM formulation is explicitly imported from prior published work (Wesel and Batselier, 2021; Batselier, 2022), and the implementation is then benchmarked on standard nonlinear system identification problems. There is no equation in which a predicted quantity is defined in terms of the target output by construction, no fitted parameter is subsequently reported as a prediction, and no uniqueness theorem from the authors' prior work is invoked to force the model choice. Hyperparameters are selected on training/validation data and the final models are evaluated on unseen test sequences, with baseline RMSE values taken from an external, non-author Nonlinear Benchmark website. While the comparability of baseline and TNKM protocols is a legitimate external-validity concern, it is not a circularity: the benchmark values are not outputs of the paper's own fitting procedure. The authors' self-citations are to prior published formulations and standard results (e.g., KRR as the unconstrained solution), and they are not used to forbid alternative interpretations. No circular step can be exhibited, so the circularity score is 0.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The model hyperparameters (tensor ranks, feature degrees, Fourier frequency counts, lag sets, regularization strengths) are selected per benchmark and are the main free choices. The mathematical framework is a re-statement of established TNKM theory (Wesel and Batselier 2021; Batselier 2022; Oseledets 2011 for TT), so no new axioms or invented entities are introduced. The assumptions listed are standard convergence and representability assumptions for the optimizers and feature maps.

free parameters (7)

CP rank R for Coupled Electric Drives = 15
CPKM rank R=15 in Section 4.2; selected via cross-validation, controls model complexity and the reported RMSEs.
CP rank R for Cascaded Tanks = 13
CPKM rank R=13 in Section 4.2; cross-validated, affects the reported RMSE of 0.347.
TT rank R for Silverbox = 4
TTKM rank R=4 in Section 4.2; cross-validated, affects Silverbox RMSE values in Table 4.
Polynomial feature degree I_d per benchmark = 3 (Coupled), 2 (Cascaded), 2 (Silverbox)
Section 4.2; chosen via cross-validation and defines the expressiveness of the feature map.
Fourier feature dimension and period for Airfoil = I_d=20; period L and regularization not fully specified in text
Section 4.1 sets I_d=20 and CP rank R=10; these choices determine the optimization comparison.
NARX lag sets I_u, I_y = Per benchmark, e.g., {0..10} and {1..10} for Coupled; {1,2,3,4,8,12,16,32} for Tanks; {0..48} and {1..48} for Silverbox
Selected heuristically in Section 4.2 from partial autocorrelation; directly sets the model input structure.
Regularization parameters = gamma_w=1e-3, beta_e=1.0 in Listing 1; benchmark values not fully reported
Regularization of tensor cores influences fit and stability; incomplete reporting makes exact reproduction harder.

assumptions (4)

standard math CP and TT decompositions can represent the model weight tensor in low-rank form, and the corresponding contractions in Equations (8)-(9) are valid.
Invoked in Section 2.3 as the basis for compact parameterization; these are standard tensor decomposition facts.
domain assumption Polynomial and Fourier feature maps are sufficiently expressive to approximate the target functions on the compact, normalized input domains.
Invoked in Section 2.2 with Stone-Weierstrass for polynomials and Fourier eigenfunction properties; limits the achievable model accuracy.
domain assumption The benchmark dynamical systems can be represented by finite-lag NARX models with the chosen lag sets, and recursive simulation remains stable.
Section 4.2 defines the model class via Equation (11) and selects lags heuristically; if the lag structure is insufficient, reported RMSE values would not reflect the framework's capabilities.
domain assumption Alternating least squares converges to a satisfactory solution of the multilinear least-squares subproblem within the stated number of epochs.
Section 2.4 relies on ALS convergence for the reported fast and accurate fits; no convergence proof specific to TNKM is given.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Tensor Network Kernel Machines: A JAX Framework for Machine Learning and Nonlinear System Identification." pith.science (2026). https://pith.science/paper/ZVC6D7B2

@misc{pith2026260807043,
  author       = {Pith},
  title        = {Pith review of: Tensor Network Kernel Machines: A JAX Framework for Machine Learning and Nonlinear System Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZVC6D7B2}},
  note         = {Machine review of arXiv:2608.07043}
}

read the original abstract

Developing nonlinear models that are both expressive and computationally efficient remains a challenge in machine learning and nonlinear system identification. Tensor network kernel machines (TNKM) address this challenge by combining nonlinear feature representations with compact low-rank tensor-network parameterizations. However, practical and extensible software frameworks for developing TNKM models remain limited. In this work, we introduce "tnkm", an open-source Python library for constructing and training TNKM models using JAX. The library provides a unified interface for combining different feature maps, tensor-network architectures, and optimization strategies, including alternating least squares and gradient-based methods. We demonstrate the capabilities of "tnkm" on nonlinear benchmark problems, showing that the implemented models achieve competitive prediction accuracy while retaining compact parameterizations and efficient training. The proposed framework facilitates reproducible development and application of tensor-network-based learning methods.

Figures

Figures reproduced from arXiv: 2608.07043 by the authors.

**Figure 2.** Architecture of the tnkm library, showing the main components used to construct and train tensor network kernel machines. while regularization of v constrains the low-rank representation and provides a computationally cheaper alternative. The tnkm library provides two complementary optimization strategies for training TNKM models: alternating least squares (ALS) and gradient-based optimization. The choice of optim… view at source ↗

**Figure 3.** Validation MSE versus elapsed training time for [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Simulation results on the Coupled Electric Drives [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

**Figure 6.** Figure 6: Simulation results on the Silverbox benchmark [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 37 canonical work pages

[1]

Wigren and M

T. Wigren and M. Schoukens. Coupled electric drives data set and reference models. 2017

work page 2017
[2]

Proceedings of the European Control Conference (ECC) , year =

Torbjörn Wigren and Johan Schoukens , title =. Proceedings of the European Control Conference (ECC) , year =

work page
[3]

Schoukens and J.P

M. Schoukens and J.P. Noël , keywords =. Three Benchmarks Addressing Open Challenges in Nonlinear System Identification , journal =. 2017 , note =

work page 2017
[4]

Pope and Michael Marcolini , title =

Thomas Brooks and D. Pope and Michael Marcolini , title =. 1989 , howpublished =

work page 1989
[5]

2017 , eprint=

Adam: A Method for Stochastic Optimization , author=. 2017 , eprint=

2017
[6]

Journal of Machine Learning Research , year =

James Hensman and Nicolas Durrande and Arno Solin , title =. Journal of Machine Learning Research , year =

work page
[7]

Hilbert space methods for reduced-rank

Solin, Arno and Särkkä, Simo , year=. Hilbert space methods for reduced-rank. Statistics and Computing , publisher=. doi:10.1007/s11222-019-09886-w , number=

doi:10.1007/s11222-019-09886-w
[8]

Large-Scale Learning with

Wesel, Frederiek and Batselier, Kim , booktitle =. Large-Scale Learning with

work page

Show all 50 references

[9]

Quantized

Wesel, Frederiek and Batselier, Kim , booktitle =. Quantized. 2024 , editor =

2024
[10]

Louis De Branges , journal =. The
[11]

A Tensor Network

Kim Batselier and Zhongming Chen and Ngai Wong , keywords =. A Tensor Network. Automatica , volume =. 2017 , issn =

2017
[12]

Cichocki, Andrzej and Lee, Namgil and Oseledets, Ivan and Phan, Anh-Huy and Zhao, Qibin and Mandic, Danilo P. , year=. Tensor Networks for Dimensionality Reduction and Large-scale Optimization: Part 1 Low-Rank Tensor Decompositions , volume=. Foundations and Trends® in Machine...

doi
[13]

and Bader, Brett W

Kolda, Tamara G. and Bader, Brett W. , title =. SIAM Review , volume =. 2009 , doi =

2009
[14]

Oseledets, I. V. , title =. SIAM Journal on Scientific Computing , volume =. 2011 , doi =

2011
[15]

Laplace Approximation for

Albert Saiapin and Kim Batselier , year=. Laplace Approximation for. 2604.26673 , archivePrefix=

arXiv
[16]

2025 , eprint=

Laplace Approximation For Tensor Train Kernel Machines In System Identification , author=. 2025 , eprint=

2025
[17]

DeepMind and Babuschkin, Igor and Baumli, Kate and Bell, Alison and Bhupatiraju, Surya and Bruce, Jake and Buchlovsky, Peter and Budden, David and Cai, Trevor and Clark, Aidan and Danihelka, Ivo and Dedieu, Antoine and Fantacci, Claudio and Godwin, Jonathan and Jones, Chris an...
[18]

James Bradbury and Roy Frostig and Peter Hawkins and Matthew James Johnson and Yash Katariya and Chris Leary and Dougal Maclaurin and George Necula and Adam Paszke and Jake Vander
[19]

ArXiv , year=

Tensor Ring Decomposition , author=. ArXiv , year=
[20]

SIAM Journal on Matrix Analysis and Applications , volume =

Grasedyck, Lars , title =. SIAM Journal on Matrix Analysis and Applications , volume =
[21]

Proceedings of The 28th International Conference on Artificial Intelligence and Statistics , pages =

Tensor Network Based Feature Learning Model , author =. Proceedings of The 28th International Conference on Artificial Intelligence and Statistics , pages =. 2025 , editor =

2025
[22]

Interpretable

Afra Kilic and Kim Batselier , year=. Interpretable. 2507.11136 , archivePrefix=

arXiv
[23]

Rasmussen, Carl Edward and Williams, Christopher K. I. , title =. 2005 , isbn =

2005
[24]

ArXiv , year=

An Introduction to Convolutional Neural Networks , author=. ArXiv , year=
[25]

2021 , eprint=

Hyperparameter Optimization: Foundations, Algorithms, Best Practices and Open Challenges , author=. 2021 , eprint=

2021
[26]

, title =

Scholkopf, Bernhard and Smola, Alexander J. , title =. 2001 , isbn =

2001
[27]

2016 , isbn =

Goodfellow, Ian and Bengio, Yoshua and Courville, Aaron , title =. 2016 , isbn =

2016
[28]

Nonlinear system identification: a user-oriented road map

Johan Schoukens and Lennart Ljung. Nonlinear system identification: a user-oriented road map. IEEE Control Systems. 2019. doi:10.1109/MCS.2019.2938121

2019
[29]

, title =

Bishop, Christopher M. , title =. 2006 , isbn =

2006
[30]

Nature , volume=

Deep learning , author=. Nature , volume=. 2015 , publisher=

2015
[31]

Low-Rank Tensor Decompositions for Nonlinear System Identification: A Tutorial with Examples , year=

Batselier, Kim , journal=. Low-Rank Tensor Decompositions for Nonlinear System Identification: A Tutorial with Examples , year=
[32]

and Trofimov, M

Novikov, A. and Trofimov, M. and Oseledets, I. , year=. Exponential machines , volume=. Bulletin of the Polish Academy of Sciences. Technical Sciences , publisher=
[33]

Tensor Networks for Dimensionality Reduction and Large-Scale Optimizations Part 2 Applications and Future Perspectives , volume=

Cichocki, Andrzej and Phan, Anh-Huy and Zhao, Qibin and Lee, Namgil and Oseledets, Ivan and Sugiyama, Masashi and Mandic, Danilo , year=. Tensor Networks for Dimensionality Reduction and Large-Scale Optimizations Part 2 Applications and Future Perspectives , volume=. Foundatio...

doi
[34]

Journal of Machine Learning Research (JMLR) , volume =

Jean Kossaifi and Yannis Panagakis and Anima Anandkumar and Maja Pantic , title =. Journal of Machine Learning Research (JMLR) , volume =
[35]

Johannes Hauschild and Jakob Unfried and Sajant Anand and Bartholomew Andrews and Marcus Bintz and Umberto Borla and Stefan Divic and Markus Drescher and Jan Geiger and Martin Hefel and Kévin Hémery and Wilhelm Kadow and Jack Kemp and Nico Kirchner and Vincent S. Liu and Gunna...

2024
[36]

2018 , publisher =

Gray, Johnnie , title =. 2018 , publisher =

2018
[37]

2025 , eprint=

tn4ml: Tensor Network Training and Customization for Machine Learning , author=. 2025 , eprint=

2025
[38]

Random Features for Large-Scale Kernel Machines , volume =

Rahimi, Ali and Recht, Benjamin , booktitle =. Random Features for Large-Scale Kernel Machines , volume =
[39]

2002 , publisher =

Least Squares Support Vector Machines , author =. 2002 , publisher =

2002
[40]

Parallelized Tensor Train Learning of Polynomial Classifiers , volume =

Chen, Zhongming and Batselier, Kim and Suykens, Johan and Wong, Ngai , year =. Parallelized Tensor Train Learning of Polynomial Classifiers , volume =
[41]

SIAM Journal on Scientific Computing , volume =

Holtz, Sebastian and Rohwedder, Thorsten and Schneider, Reinhold , title =. SIAM Journal on Scientific Computing , volume =
[42]

and Luciani, X

Comon, P. and Luciani, X. and de Almeida, A. L. F. , title =. Journal of Chemometrics , volume =
[43]

Large-Scale Machine Learning with Stochastic Gradient Descent

Bottou, L \'e on. Large-Scale Machine Learning with Stochastic Gradient Descent. Proceedings of COMPSTAT'2010. 2010

2010
[44]

2014 , eprint=

Generating Sequences With Recurrent Neural Networks , author=. 2014 , eprint=

2014
[45]

International Journal of Adaptive Control and Signal Processing , volume =

Forgione, Marco and Piga, Dario , title =. International Journal of Adaptive Control and Signal Processing , volume =
[46]

Beintema and Maarten Schoukens and Roland T \'o th

G.I. Beintema and Maarten Schoukens and Roland T \'o th. Continuous-time identification of dynamic state-space models by deep subspace encoding. The Eleventh International Conference on Learning Representations. 2023

2023
[47]

On the normalization of gray-box state-space neural networks , journal =

Milad Banitalebi Dehkordi and Marco Forgione and Gionata Cimini and Lapo Frascati and Dario Piga , keywords =. On the normalization of gray-box state-space neural networks , journal =. 2026 , issn =

2026
[48]

2026 , eprint=

Bendegúz Györök and Roel Drenth and Chris Verhoek and Tamás Péni and Maarten Schoukens and Roland Tóth , title=. 2026 , eprint=

2026
[49]

Proceedings of the 3rd Conference on Learning for Dynamics and Control , pages =

Nonlinear state-space identification using deep encoder networks , author =. Proceedings of the 3rd Conference on Learning for Dynamics and Control , pages =. 2021 , publisher =

2021
[50]

Inference and Learning of Nonlinear LFR State-Space Models , year=

Floren, Merijn and Noël, Jean-Philippe and Swevers, Jan , journal=. Inference and Learning of Nonlinear LFR State-Space Models , year=

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Wigren and M

T. Wigren and M. Schoukens. Coupled electric drives data set and reference models. 2017

work page 2017

[2] [2]

Proceedings of the European Control Conference (ECC) , year =

Torbjörn Wigren and Johan Schoukens , title =. Proceedings of the European Control Conference (ECC) , year =

work page

[3] [3]

Schoukens and J.P

M. Schoukens and J.P. Noël , keywords =. Three Benchmarks Addressing Open Challenges in Nonlinear System Identification , journal =. 2017 , note =

work page 2017

[4] [4]

Pope and Michael Marcolini , title =

Thomas Brooks and D. Pope and Michael Marcolini , title =. 1989 , howpublished =

work page 1989

[5] [5]

2017 , eprint=

Adam: A Method for Stochastic Optimization , author=. 2017 , eprint=

2017

[6] [6]

Journal of Machine Learning Research , year =

James Hensman and Nicolas Durrande and Arno Solin , title =. Journal of Machine Learning Research , year =

work page

[7] [7]

Hilbert space methods for reduced-rank

Solin, Arno and Särkkä, Simo , year=. Hilbert space methods for reduced-rank. Statistics and Computing , publisher=. doi:10.1007/s11222-019-09886-w , number=

doi:10.1007/s11222-019-09886-w

[8] [8]

Large-Scale Learning with

Wesel, Frederiek and Batselier, Kim , booktitle =. Large-Scale Learning with

work page