Pith. sign in

REVIEW 2 major objections 6 minor 62 references

An Efficient and Interpretable Autoregressive Model for High-Dimensional Tensor-Valued Time Series

T0 review · 2 major / 6 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 The paper claims that a CP-based factorization of the coefficient tensor makes tensor autoregression interpretable as supervised factor regression while keeping parameters linear in tensor order.

desk verdict A genuinely new two-sided CP structure for tensor autoregression, worth refereeing despite a real but likely patchable gap in the covering-number proofs. read the letter →

arxiv 2506.01658 v1 pith:3JNZ4ZF4 submitted 2025-06-02 stat.ME

Yuxi Cai , Lan Li , Yize Wang , Guodong Li This is my paper · ORCID
classification stat.ME MSC 62M1062J0715A69
keywords tensor-valuedtimeseriesautoregressionCPdecompositionsupervisedfactormodellow-rankplussparsenonasymptoticboundsalternatingleastsquaresElNiño-SouthernOscillation
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-valued time series are hard to model because the coefficient tensor of even an order-one autoregression has twice the data dimension, making direct estimation hopeless. The paper proposes a middle path between Tucker and CP decompositions: it factors the coefficient tensor so that response features extracted by one set of CP loadings are regressed on covariate features extracted by another set. This supervised-factor structure keeps the parameter count linear in tensor order, specifically $P R_y R_x + (R_y + R_x) \sum_i q_i$ for an AR($P$) model, and the paper derives nonasymptotic ordinary-least-squares error bounds that match this complexity. A low-rank-plus-sparse extension is added to absorb heterogeneous or spiky signals that an exact low-rank assumption would miss. The motivating application is forecasting ocean temperature anomalies related to El Niño, where the fitted loadings recover the El Niño basin.

What carries the argument

The load-bearing object is the decomposition in equations (2.4)-(2.5), $[A]_n = \Lambda_y (\Lambda_y^\top \Lambda_y)^{-1} G \Lambda_x^\top$, which rewrites tensor-on-tensor regression as a regression of extracted response features on extracted covariate features. The loading matrices are Khatri-Rao products of unit-norm factor columns, giving each column the form of a vectorized CP rank-one tensor, and in the autoregressive version the same $\Lambda_y$ and $\Lambda_x$ are shared across lags with only $G_k$ varying. The proofs use covering-number bounds for the difference set of two such coefficient tensors together with restricted strong convexity of the empirical loss, and the optimization is an alternating least squares routine in which each block update reduces to a closed-form linear regression.

What would settle it

Simulate a stationary tensor AR(1) model with moderate dimensions, choose loading matrices whose columns are nearly collinear (angle around one degree), and compare the OLS estimation error against the predicted $\sqrt{d_{\mathrm{AR}} \log(\cdot)/T}$ rate as the angle decreases; if the error grows substantially, the rate is carried by an unstated uniform representation-norm condition rather than the assumptions listed.

Watch

Extended reading notes

Core claim

The central claim is that the coefficient tensor of a tensor-on-tensor regression or tensor autoregression can be parameterized, via sequential matricization, as $[A]_n = \Lambda_y (\Lambda_y^\top \Lambda_y)^{-1} G \Lambda_x^\top$, where $\Lambda_y$ and $\Lambda_x$ are Khatri-Rao products of unit-norm loading matrices. Multiplying the regression equation by $\Lambda_y^\top$ yields $f_t^{\mathrm{response}} = G f_t^{\mathrm{covariate}} + e_t^{\mathrm{CP}}$, so the model literally regresses response features on covariate features. The paper argues that this CP-based low-rank structure combines the interpretability of supervised factor models with the parsimony of CP decomposition, and that the OLS estimator over this parameter space attains Frobenius-error bounds of order $\sqrt{d_{\mathrm{AR}} d_c / (T-P)}$ under stationarity and sub-Gaussian noise. For the low-rank-plus-sparse variant, the error splits into an estimation term and a sample-size-independent unidentifiability term, the latter governed by the radius $\alpha_L$.

Load-bearing premise

The theorems rely on every coefficient tensor in the parameter space having a representation whose inner matrix $G$ is bounded in norm; the paper only assumes non-degeneracy of the loadings, which excludes exact linear dependence but permits nearly collinear columns where that norm can diverge.

Editorial extensions

If this is right

  • For an AR($P$) model the parameter count is $d_{\mathrm{AR}} = P R_y R_x + (R_y + R_x) \sum_i q_i$, so the number of parameters grows linearly with tensor order instead of exponentially as in Tucker-based tensor AR.
  • The OLS estimator satisfies $\|\hat{A} - A^*\|_F \lesssim (\kappa^2 \kappa'_{U,B} / \kappa_{L,A}) \sqrt{d_{\mathrm{AR}} d_c / (T-P)}$ with high probability, making the rate explicit in the sample size, ranks, and dimensions.
  • The low-rank-plus-sparse estimator separates low-rank collaborative structure from sparse idiosyncratic structure, with squared error bounded by $(d_{\mathrm{AR}} d_c + s \log(P Q^2))/(T-P)$ plus an unidentifiability term $s \alpha_L^2/(P Q^2)$ that does not vanish with sample size.
  • The fitted response loadings from the ENSO example concentrate weight on the El Niño basin, giving the model a concrete spatial interpretation that pure CP or Lasso estimates lack.

Reading between the lines

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

  • The same factor-regression parametrization applies to any tensor-on-tensor regression, so the interpretability-efficiency trade-off it resolves is not specific to time series; one could test it directly on cross-sectional imaging or marketing data.
  • Sharing $\Lambda_y$ and $\Lambda_x$ across lags presumes a single linear feature extractor for all temporal dynamics; regime-switching or time-varying dynamics would likely require either multiple factor sets or an extension where loadings evolve.
  • The theorem's unidentifiability term suggests that any practical use of the low-rank-plus-sparse model should report the chosen $\alpha_L$ together with the fitted sparse tensor, since the same likelihood can be explained by shifting mass between the two components.
  • A natural stress test of the theory is to let the loading columns become progressively closer to collinear and check whether the empirical error degrades as the covering-number argument predicts; the paper's assumptions do not rule out such a degradation.
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.

Referee Report

2 major / 6 minor

Summary. This paper proposes a new low-rank structure for the coefficient tensor in tensor-on-tensor regression and tensor autoregression. The structure writes the sequential matricization of the coefficient as Λ_y G Λ_xᵀ (with a Kronecker factor for lags), so that response and covariate features are extracted by CP-type loadings and linked by a small coefficient matrix G. This yields a parameter count linear in tensor order and a supervised-factor interpretation. The paper also adds a sparse component (low-rank plus sparse) with an identifiability radius α_L, gives OLS and penalized estimators, an alternating least-squares algorithm, nonasymptotic bounds (Theorems 1–4), simulations, and an ENSO application.

Significance. If the rate results are correct, the paper's contribution is substantial: it provides an interpretable alternative to Tucker-based tensor AR with dramatically smaller parameter count, and the low-rank-plus-sparse extension addresses misspecification. The paper is careful to state the sample-size-independent unidentifiability error in Theorem 4, and the supplement contains detailed proofs. The main obstacle is the missing uniform conditioning assumption in the covering-number arguments; this is a fixable gap but is load-bearing for the stated theorems.

major comments (2)
  1. [Section B, Lemma 1(b)–(d), used in Theorems 1–4] The proofs of the covering-number bounds bound the middle factor by 'By Assumption 3, ∥G_1∥_F ≤ ... and same for G_2' for arbitrary A_1, A_2 in the parameter space. Assumption 3, however, bounds only the G of the true coefficient tensor, not all elements of Θ. Since S(q,R) only constrains columns to be unit-norm and does not bound singular values of Λ_y or Λ_x away from zero, there exist elements of Θ with ∥A∥_F ≤ 1 for which every representation requires arbitrarily large ∥G∥_F (nearly collinear Khatri–Rao columns). The nets in (B.1) and the Gaussian-width estimate (B.14) therefore do not cover the difference set Ξ', and the RSC/deviation lemmas are not justified. Please add a uniform conditioning assumption (e.g., σ_min(Λ_y), σ_min(Λ_x) ≥ c > 0 for all admissible loadings, or an explicit bound on the minimal representation norm of G) and verify that Lemmas 1–7 and Theorems 1–4 follow.
  2. [Eq. (2.4) and Section 2.3] The decomposition uses (Λ_yᵀΛ_y)⁻¹, so Λ_y must have full column rank for the model to be well defined. The 'non-degeneracy' condition stated after Eq. (2.3) is not incorporated into the formal parameter space S(q,R) or into Assumptions 1–7; moreover, after the reparameterization G := (Λ_yᵀΛ_y)⁻¹G, the quantity bounded in Assumption 3 is not the original G. Please make the full-rank/quantitative conditioning requirement a formal assumption; this also helps resolve the issue in the previous comment.
minor comments (6)
  1. [Section B, Lemma 1(b)] The proof defines g′ with an extra factor P relative to the statement of the lemma; please reconcile the two definitions so the constants in the covering bound are consistent.
  2. [Assumption 1 and Theorems 1–3] Because c_x and C_x are allowed to depend on dimensions and to shrink or diverge, the phrases 'bounded away from zero and infinity' in the discussion after Theorems 1–3 need a precise convention; otherwise the displayed rates hide dimension-dependent constants.
  3. [Theorem 4] The statement lists only Assumptions 3–5 in its hypotheses; include Assumptions 6 and 7 explicitly for completeness.
  4. [Algorithm 2] The update of A_L via Algorithm 1 does not enforce the constraint ∥A_L∥_∞ ≤ α_L/(P Q²) from (2.11); clarify whether a final trimming step is applied or whether the constrained subproblem is solved.
  5. [Section 5, Table 2] The text abbreviates the error as MAFE while the table header reads MSAE; make the abbreviation consistent.
  6. [Section B, Lemma 1(d)] The notation ∥δ∥_2 = ∥∆^j∥_2 refers to an undefined ∆^j; define this row quantity explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the proposed decomposition is an explicit modeling assumption, and the non-asymptotic bounds are derived from stated assumptions rather than from fitted constants or load-bearing self-citations.

full rationale

The central structural step, equations (2.4)-(2.5), is introduced as an assumption ('we propose to assume that the sequential matricization of A has the following decomposition'), not as a prediction derived from the data. The reformulation f_response_t = G f_covariate_t + e_t^CP is exactly the same model after multiplying by Lambda_y^T, so it is a reparameterization, not a circular derivation. Theorems 1-4 are proved in the supplement via generic chaining, restricted strong convexity, and covering-number arguments under Assumptions 1-7; no fitted constant is recycled as a prediction, and the unidentifiability error in Theorem 4 is explicitly acknowledged as sample-size-independent. Self-citations (Wang et al. 2024; Huang et al. 2025; Si et al. 2024; Cai et al. 2022, 2023) are used for context, baselines, and algorithmic ideas, but none is load-bearing for the main derivation. The simulation and ENSO experiments use hold-out validation and test-set evaluation, which is standard and not circular. One technical gap exists: Lemma 1(b) bounds G1 and G2 using Assumption 3's constant g, but that assumption constrains only the true coefficient's G; arbitrary elements of Theta need not satisfy ||G||_op <= g without a uniform well-conditioning condition on Lambda_y and Lambda_x. This is a proof gap and a correctness risk, not a circularity, because it does not make any theorem's conclusion an input by construction.

Assumptions & free parameters 4 free parameters · 9 assumptions · 0 invented entities

The central claim rests on statistical assumptions 1-7 (sub-Gaussian inputs and errors, bounded G, stationarity, identifiability radius, sparsity level), the explicit non-degeneracy of the loading towers, and an implicit well-conditioning premise needed for the covering arguments. Free parameters are the tuning choices (ranks, order, penalty lambda) and the hand-set radius alpha_L. No invented entities are introduced.

free parameters (4)
  • Radius of non-identifiability alpha_L = P^{1/2} Q, hand-chosen for simplicity
    Sets the trimming threshold alpha_L/(P Q^2) in Algorithm 2 and the unidentifiability error s alpha_L^2/(P Q^2) in Theorem 4; it is not estimated from data (Section 2.4, Remark 4).
  • Autoregressive order P = 1 in the ENSO application
    Selected by hold-out rolling forecast over 1 <= P <= 12 in the ENSO example; a standard model-selection choice listed for completeness.
  • Ranks R_y and R_x = R_y = 3, R_x = 2 in the ENSO application
    Selected by hold-out for the low-rank model and reused without re-selection for the low-rank plus sparse model, a minor inconsistency noted in Section 5.
  • Lasso penalty lambda for the sparse part = 0.01 in the ENSO application
    Selected by hold-out from the candidate grid in Section 5; a standard tuning parameter.
assumptions (9)
  • domain assumption Covariates are i.i.d., zero-mean, sigma^2-sub-Gaussian with spectrum bounded as c_x <= lambda_min(Sigma_x) <= lambda_max(Sigma_x) <= C_x (Assumption 1).
    Invoked in Lemmas 2, 3 and 8 to obtain restricted strong convexity and the deviation bound for the i.i.d. regression setting of Theorem 1.
  • domain assumption Errors are i.i.d., mean-zero and kappa^2-sub-Gaussian conditional on covariates (Assumption 2).
    Used for the empirical process deviation bound in Lemma 3.
  • domain assumption The coupling matrix G has operator norm bounded by a constant g (Assumption 3).
    Bounded G is used in the covering-number estimates (Lemma 1) via g' = g sqrt(min(R_y, P R_x)); it is stated for the true parameter only, which is part of the gap flagged in the weakest assumption.
  • domain assumption Autoregression errors satisfy e_t = Sigma_e^{1/2} xi_t with independent kappa^2-sub-Gaussian entries and bounded spectrum (Assumption 4).
    Needed for the Hanson-Wright and martingale concentration steps in Lemmas 4, 5, 10 and 11.
  • domain assumption Stationarity: det(I_Q - A_1 z - ... - A_P z^P) != 0 for |z| < 1 (Assumption 5).
    Guarantees the VMA representation and the spectral quantities mu_min and mu_max from Basu and Michailidis (2015) used throughout the AR proofs.
  • domain assumption Identifiability of the low-rank and sparse parts: the true low-rank coefficient satisfies ||A_L||_infinity <= alpha_L/(P Q^2) (Assumption 6).
    Without it the low-rank and sparse components are not separable; it also generates the sample-size-independent unidentifiability error in Theorem 4.
  • domain assumption The true sparse tensor A_S has at most s nonzero entries (Assumption 7).
    Used in Lemmas 6 and 7 via the sparse covering bound with log(P Q^2) factors.
  • domain assumption The CP loading towers are non-degenerate: a linear combination of the rank-one loading tensors equals zero only if all coefficients are zero, so Lambda_y^T Lambda_y is invertible (Section 2.1, after eq. 2.3).
    Required for the parametrization [A]_n = Lambda_y (Lambda_y^T Lambda_y)^{-1} G Lambda_x^T and for the feature-on-feature reading; it is not restated among Assumptions 1-7 in Section 3.
  • ad hoc to paper An implicit uniform well-conditioning premise: every element of Theta admits a representation with ||G||_F bounded by g sqrt(min(R_y, P R_x)), or equivalently the Khatri-Rao loadings have singular values uniformly bounded away from zero.
    The covering nets in Lemma 1(b)-(d) only cover bounded-G representations; near-degenerate loadings are not excluded by the stated assumptions, so the difference set Xi may not be covered. This premise is load-bearing for Theorems 1-4 but is never stated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Efficient and Interpretable Autoregressive Model for High-Dimensional Tensor-Valued Time Series." pith.science (2026). https://pith.science/paper/3JNZ4ZF4

@misc{pith2026250601658,
  author       = {Pith},
  title        = {Pith review of: An Efficient and Interpretable Autoregressive Model for High-Dimensional Tensor-Valued Time Series},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3JNZ4ZF4}},
  note         = {Machine review of arXiv:2506.01658}
}
abstract

In autoregressive modeling for tensor-valued time series, Tucker decomposition, when applied to the coefficient tensor, provides a clear interpretation of supervised factor modeling but loses its efficiency rapidly with increasing tensor order. Conversely, canonical polyadic (CP) decomposition maintains efficiency but lacks a precise statistical interpretation. To attain both interpretability and powerful dimension reduction, this paper proposes a novel approach under the supervised factor modeling paradigm, which first uses CP decomposition to extract response and covariate features separately and then regresses response features on covariate ones. This leads to a new CP-based low-rank structure for the coefficient tensor. Furthermore, to address heterogeneous signals or potential model misspecifications arising from stringent low-rank assumptions, a low-rank plus sparse model is introduced by incorporating an additional sparse coefficient tensor. Nonasymptotic properties are established for the ordinary least squares estimators, and an alternating least squares algorithm is introduced for optimization. Theoretical properties of the proposed methodology are validated by simulation studies, and its enhanced prediction performance and interpretability are demonstrated by the El Ni$\tilde{\text{n}}$o-Southern Oscillation example.

Figures

Figures reproduced from arXiv: 2506.01658 by the authors.

Figure 1
Figure 1. The grid points represent the geographical locations where ocean temperature data [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Average estimation errors ∥Ab − A∥F for the tensor autoregression model with low-rank structure under settings: (a) changing the sample size T, (b) changing the ranks Ry and Rx, and (c) changing the dimensions qi ’s. Three different ways of generating the error Et are considered, as specified in the legend. 4 Simulation studies This section conducts two experiments to evaluate finite-sample performances of the pro￾p… view at source ↗
Figure 3
Figure 3. Average errors ∥Ae L − AL∥ 2 F + ∥Ae S − AS∥ 2 F for tensor autoregression model with low-rank plus sparse structure under settings: (a) changing the sample size T, (b) changing the ranks Ry and Rx, and (c) changing the dimensions qi ’s. Three different ways of generating the error Et are considered, as specified in the legend. From Theorem 2, the estimation error converges at a rate of p dARdc/(T − P), where dAR = … view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Heatmaps of spatial loadings for response features: the top and bottom panels are [PITH_FULL_IMAGE:figures/full_fig_p027_4.png]
Figure 5
Figure 5. Figure 5: Heatmaps of variable loadings for response features: the left and right panels are [PITH_FULL_IMAGE:figures/full_fig_p028_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 52 canonical work pages

  1. [1]

    Agarwal, A., Negahban, S., and Wainwright, M. J. (2012). Noisy matrix decomposition via convex relaxation: Optimal rates in high dimensions. The Annals of statistics , 40(2):1171--1197

    work page 2012

  2. [2]

    T., Yu, Q

    Bahadori, M. T., Yu, Q. R., and Liu, Y. (2014). Fast multivariate spatio-temporal analysis via low rank tensor learning. In Ghahramani, Z., Welling, M., Cortes, C., Lawrence, N., and Weinberger, K., editors, Advances in Neural Information Processing Systems , volume 27. Curran Associates, Inc

    work page 2014

  3. [3]

    and Wang, P

    Bai, J. and Wang, P. (2016). Econometric Analysis of Large Factor Models . Annual Review of Economics , 8(1):53--80

    work page 2016

  4. [4]

    Bai, P., Safikhani, A., and Michailidis, G. (2023). Multiple change point detection in reduced rank high dimensional vector autoregressive models. Journal of the American Statistical Association , 118(544):2776--2792

    work page 2023

  5. [5]

    Banerjee, A., Chen, S., Fazayeli, F., and Sivakumar, V. (2015). Estimation with norm regularization

    work page 2015

  6. [6]

    Basu, S., Li, X., and Michailidis, G. (2019). Low rank and structured modeling of high-dimensional vector autoregressions. IEEE Transactions on Signal Processing , 67(5):1207--1222

    work page 2019

  7. [7]

    and Michailidis, G

    Basu, S. and Michailidis, G. (2015). Regularized estimation in sparse high-dimensional time series models. The Annals of Statistics , 43:1535--1567

    2015

  8. [8]

    and Parys, B

    Bertsimas, D. and Parys, B. V. (2020). Sparse high-dimensional regression: Exact scalable algorithms and phase transitions. The Annals of Statistics , 48(1):pp. 300--323

    work page 2020

Show all 62 references
  1. [9]

    Bi, X., Qu, A., and Shen, X. (2018). Multilayer tensor factorization with applications to recommender systems. The Annals of Statistics , 46(6B):3308--3333

    2018

  2. [10]

    Billio, M., Casarin, R., and Iacopini, M. (2024). Bayesian markov-switching tensor regression for time-varying networks. Journal of the American Statistical Association , 119(545):109--121

    2024

  3. [11]

    V., and Chen, Y

    Cai, C., Poor, H. V., and Chen, Y. (2020). Uncertainty quantification for nonconvex tensor completion: Confidence intervals, heteroscedasticity and optimality. In III, H. D. and Singh, A., editors, Proceedings of the 37th International Conference on Machine Learning , volume 1...

    2020

  4. [12]

    Cai, J.-F., Li, J., and Xia, D. (2022). Provable tensor-train format tensor completion by riemannian optimization. Journal of Machine Learning Research , 23(123):1--77

    2022

  5. [13]

    Cai, J.-F., Li, J., and Xia, D. (2023). Generalized low-rank plus sparse tensor estimation by fast riemannian optimization. Journal of the American Statistical Association , 118(544):2588--2604

    2023

  6. [14]

    Chen, R., Yang, D., and Zhang, C.-H. (2022). Factor models for high-dimensional tensor time series. Journal of the American Statistical Association , 117(537):94--116

    2022

  7. [15]

    Clarke, A. J. and Van Gorder, S. (2003). Improving el niño prediction using a space-time integration of indo-pacific winds and equatorial pacific upper ocean heat content. Geophysical Research Letters , 30(7)

    2003

  8. [16]

    A., Zang, P., and Zheng, T

    Davis, R. A., Zang, P., and Zheng, T. (2016). Sparse vector autoregressive modeling. Journal of Computational and Graphical Statistics , 25(4):1077--1096

    2016

  9. [17]

    W., and Takane, Y

    de Leeuw, J., Young, F. W., and Takane, Y. (1976). Additive structure in qualitative data: An alternating least squares method with optimal scaling features. Psychometrika , 41(4):471--503

    1976

  10. [18]

    Dirksen, S. (2015). Tail bounds via generic chaining . Electronic Journal of Probability , 20:1 -- 29

    2015

  11. [19]

    and Li, R

    Fan, J. and Li, R. (2001). Variable selection via nonconcave penalized likelihood and its oracle properties. Journal of the American Statistical Association , 96(456):1348--1360

    2001

  12. [20]

    Feng, L., Bi, X., and Zhang, H. (2021). Brain regions identified as being associated with verbal reasoning through the use of imaging regression via internal variation. Journal of the American Statistical Association , 116(533):144--158. PMID: 34955572

    2021

  13. [21]

    R., Yan, H., Paynabar, K., and Shi, J

    Gahrooei, M. R., Yan, H., Paynabar, K., and Shi, J. (2021). Multiple tensor-on-tensor regression: An approach for modeling processes with heterogeneous sources of data. Technometrics , 63(2):147--159

    2021

  14. [22]

    Gozolchiani, A., Havlin, S., and Yamasaki, K. (2011). Emergence of el ni\ no as an autonomous component in the climate network. Phys. Rev. Lett. , 107:148501

    2011

  15. [23]

    Guhaniyogi, R., Qamar, S., and Dunson, D. B. (2017). Bayesian tensor regression. Journal of Machine Learning Research , 18(79):1--31

    2017

  16. [24]

    Guo, W., Kotsia, I., and Patras, I. (2012). Tensor learning for regression. IEEE Transactions on Image Processing , 21(2):816--827

    2012

  17. [25]

    Han, R., Willett, R., and Zhang, A. R. (2022). An optimal statistical and computational framework for generalized tensor estimation . The Annals of Statistics , 50(1):1 -- 29

    2022

  18. [26]

    Han, Y., Yang, D., Zhang, C.-H., and Chen, R. (2024). Cp factor model for dynamic tensors

    2024

  19. [27]

    Hao, B., Wang, B., Wang, P., Zhang, J., Yang, J., and Sun, W. W. (2021). Sparse tensor additive regression. Journal of Machine Learning Research , 22(64):1--43

    2021

  20. [28]

    explanatory

    Harshman, R. A. (1970). F oundations of the P A R A F A C procedure: M odels and conditions for an "explanatory" multi-modal factor analysis. UCLA Working Papers in Phonetics , 16:1--84

    1970

  21. [29]

    Hastie, T., Tibshirani, R., and Wainwright, M. (2015). Statistical Learning with Sparsity: The Lasso and Generalizations . Chapman & Hall/CRC

    2015

  22. [30]

    Huang, F., Lu, K., Zheng, Y., and Li, G. (2025). Supervised factor modeling for high-dimensional linear time series

    2025

  23. [31]

    Kolda, T. G. and Bader, B. W. (2009). Tensor decompositions and applications. SIAM Review , 51(3):455--500

    2009

  24. [32]

    and Zhang, H

    Li, C. and Zhang, H. (2021). Tensor quantile regression with application to association between neuroimages and human intelligence . The Annals of Applied Statistics , 15(3):1455 -- 1477

    2021

  25. [33]

    and Zhang, X

    Li, L. and Zhang, X. (2017). Parsimonious Tensor Response Regression . Journal of the American Statistical Association , 112(519):1131--1146

    2017

  26. [34]

    Li, X., Xu, D., Zhou, H., and Li, L. (2018). Tucker tensor regression and neuroimaging analysis. Statistics in biosciences , 10(3):520--545

    2018

  27. [35]

    and Xiao, H

    Li, Z. and Xiao, H. (2021). Multi-linear tensor autoregressive models

    2021

  28. [36]

    and Maitra, R

    Llosa-Vite, C. and Maitra, R. (2023). Reduced-rank tensor-on-tensor regression and tensor-variate analysis of variance. IEEE Transactions on Pattern Analysis and Machine Intelligence , 45(2):2282--2296

    2023

  29. [37]

    Lock, E. F. (2018). Tensor-on-tensor regression. Journal of Computational and Graphical Statistics , 27(3):638--647. PMID: 30337798

    2018

  30. [38]

    C., Abe, N., Liu, Y., and Rosset, S

    Lozano, A. C., Abe, N., Liu, Y., and Rosset, S. (2009). Grouped graphical granger modeling for gene expression regulatory networks discovery. Bioinformatics , 25(12):i110--i118

    2009

  31. [39]

    I., Bunde, A., Havlin, S., and Schellnhuber, H

    Ludescher, J., Gozolchiani, A., Bogachev, M. I., Bunde, A., Havlin, S., and Schellnhuber, H. J. (2013). Improved el niño forecasting by cooperativity detection. Proceedings of the National Academy of Sciences , 110(29):11742--11745

    2013

  32. [40]

    McPhaden, M. J. (2003). Tropical pacific ocean heat content variations and enso persistence barriers. Geophysical Research Letters , 30(9)

    2003

  33. [41]

    and Banerjee, A

    Melnyk, I. and Banerjee, A. (2016). Estimating structured vector autoregressive models. In Proceedings of the 33rd International Conference on International Conference on Machine Learning - Volume 48 , ICML'16, page 830–839. JMLR.org

    2016

  34. [42]

    Oseledets, I. V. (2011). Tensor-train decomposition. SIAM Journal on Scientific Computing , 33(5):2295--2317

    2011

  35. [43]

    Raskutti, G., Yuan, M., and Chen, H. (2019). Convex regularization for high-dimensional multiresponse tensor regression. The Annals of Statistics , 47(3):pp. 1554--1584

    2019

  36. [44]

    Sarachik, E. S. and Cane, M. A. (2010). The El Niño-Southern Oscillation Phenomenon . Cambridge University Press

    2010

  37. [45]

    Si, Y., Zhang, Y., Cai, Y., Liu, C., and Li, G. (2024). An efficient tensor regression for high-dimensional data

    2024

  38. [46]

    and Watson, M

    Stock, J. and Watson, M. (2011). Dynamic Factor Models . Oxford University Press, Oxford

    2011

  39. [47]

    Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological) , 58(1):267--288

    1996

  40. [48]

    Tucker, L. (1966). Some mathematical notes on three-mode factor analysis. Psychometrika , 31(3):279--311

    1966

  41. [49]

    Vershynin, R. (2018). High-Dimensional Probability: An Introduction with Applications in Data Science . Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press

    2018

  42. [50]

    Wainwright, M. J. (2019). High-Dimensional Statistics: A Non-Asymptotic Viewpoint . Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press

    2019

  43. [51]

    Wang, D., Zheng, Y., and Li, G. (2024). High-dimensional low-rank tensor autoregressive time series modeling. Journal of Econometrics , 238(1):105544

    2024

  44. [52]

    Wang, D., Zheng, Y., Lian, H., and Li, G. (2022). High-dimensional vector autoregressive time series modeling via tensor decomposition. Journal of the American Statistical Association , 117(539):1338--1356

    2022

  45. [53]

    Wilms, I., Basu, S., Bien, J., and Matteson, D. S. (2023). Sparse identification and estimation of large-scale vector autoregressive moving averages. Journal of the American Statistical Association , 118:571--582

    2023

  46. [54]

    R., and Zhou, Y

    Xia, D., Zhang, A. R., and Zhou, Y. (2022). Inference for low-rank tensors—no need to debias . The Annals of Statistics , 50(2):1220 -- 1245

    2022

  47. [55]

    Xu, J., Liu, X., Wilson, T., Tan, P.-N., Hatami, P., and Luo, L. (2018). Muscat: Multi-scale spatio-temporal learning with application to climate modeling. In IJCAI , pages 2912--2918

    2018

  48. [56]

    and Liu, Y

    Yu, R. and Liu, Y. (2016). Learning from multiway data: Simple and efficient tensor regression. In Balcan, M. F. and Weinberger, K. Q., editors, Proceedings of The 33rd International Conference on Machine Learning , volume 48 of Proceedings of Machine Learning Research , pages...

    2016

  49. [57]

    Zhang, X., Li, L., Zhou, H., Zhou, Y., and Shen, D. (2019). Tensor generalized estimating equations for longitudinal imaging analysis. Statistica Sinica , 29(4):1977--2005

    2019

  50. [58]

    Zhao, Q., Zhou, G., Xie, S., Zhang, L., and Cichocki, A. (2016). Tensor ring decomposition. CoRR , abs/1606.05535

    2016 arXiv

  51. [59]

    Zheng, Y. (2024). An interpretable and efficient infinite-order vector autoregressive model for high-dimensional time series. Journal of the American Statistical Association , page To appear

    2024

  52. [60]

    and Cheng, G

    Zheng, Y. and Cheng, G. (2020). Finite-time analysis of vector autoregressive models under linear restrictions . Biometrika , 108(2):469--489

    2020

  53. [61]

    Zhou, H., Li, L., and Zhu, H. (2013). Tensor regression with applications in neuroimaging data analysis. Journal of the American Statistical Association , 108(502):540--552

    2013

  54. [62]

    and Zhang, R.-H

    Zhou, L. and Zhang, R.-H. (2023). A self-attention–based neural network for three-dimensional multivariate modeling and its skillful enso predictions. Science Advances , 9(10):eadf2827

    2023

Pith tools

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