Smart Predict-then-Optimize Method with Dependent Data: Risk Bounds and Calibration of Autoregression

Chongrong Fang; Jianping He; Jixian Liu; Tao Xu

REVIEW 4 major objections 7 minor 30 references

Smart Predict-then-Optimize Method with Dependent Data: Risk Bounds and Calibration of Autoregression

T0 review · 4 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper establishes uniform calibration of the SPO+ surrogate loss for fixed-memory autoregressive predictors on stationary beta-mixing data, with explicit excess-risk rates of $O(m^{-1/4})$ for polyhedral feasible sets and…

desk verdict The SPO generalization bound under beta-mixing is a legitimate extension, but the calibration theorems for SPO+ do not follow from the proof as written because the SPO+ loss is unbounded on the paper's hypothesis class. read the letter →

arxiv 2411.12653 v1 pith:ABR4IBQ7 submitted 2024-11-19 eess.SY cs.SYstat.ML

Jixian Liu , Tao Xu , Jianping He , Chongrong Fang This is my paper · ORCID

classification eess.SYcs.SYstat.ML MSC 62M1090C1568T05

keywords predict-then-optimizeSPOlossSPO+autoregressionbeta-mixinggeneralizationboundsuniformcalibrationtimeseriesforecasting

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 extend the smart predict-then-optimize (SPO) guarantee from independent training data to dependent time series. It proposes a fixed-memory autoregressive predictor trained with the SPO+ surrogate loss and proves that, for stationary $\beta$-mixing cost sequences, the excess SPO risk of the empirical SPO+ minimizer is bounded by $C\sqrt{\log(1/\delta')}/m^{1/4}$ for bounded polyhedral feasible sets and by $C\sqrt{\log(1/\delta')}/m^{1/2}$ for strongly convex level sets. If correct, this means decision-focused training is theoretically justified for load forecasting and other dependent-data optimization tasks, not only for i.i.d. samples. The argument works by blocking the trajectory into nearly independent pieces and importing Rademacher-complexity bounds through the mixing coefficient.

What carries the argument

The load-bearing mechanism is the independent-block construction: the trajectory is split into two interleaved groups of $m$ blocks of length $a$, separated by a lag $l$, so that any two blocks are nearly independent. A coupling lemma bounds the probability error between the dependent blocks and independent blocks by $(m-1)\beta(a-l)$. This reduces the dependent-data generalization problem to an i.i.d. Rademacher-complexity bound on the blocked sample. The fixed-memory hypothesis class is the set of linear maps $f(y_{i-l:i-1})=\sum_{j=1}^{l} M_j y_{i-j}$, which makes the predictor a matrix $M$ trained by stochastic subgradient descent using a known subgradient of the SPO+ loss.

What would settle it

Construct a stationary $\beta$-mixing cost sequence whose conditional expectation is not linear in the past $l$ lags, train the fixed-memory SPO+ autoregressive model without restricting predictions, and test whether the excess SPO risk obeys the claimed $O(m^{-1/4})$ bound; violating the bound for large-magnitude predictions would confirm that the missing bounded-loss and realizability premises are carrying the argument.

Watch

Extended reading notes

Core claim

The central discovery is a uniform calibration theorem for dependent data: when cost vectors follow a stationary $\beta$-mixing process and the predictor class is a fixed-lag linear autoregression, minimizing empirical SPO+ risk controls SPO risk with high probability. Theorem 2 states this for bounded polyhedral feasible sets at rate $O(m^{-1/4})$, and Theorem 3 improves the rate to $O(m^{-1/2})$ when the feasible set is a strongly convex level set. The confidence terms degrade by an additive $2m\beta(a-l)$ factor, which is the price of dependence; when the mixing coefficient vanishes, the bounds approach the i.i.d. calibration results that the paper extends. Theorem 1 supplies the companion generalization bound for the SPO loss itself using independent blocks drawn from the dependent trajectory.

Load-bearing premise

The chain of bounds needs the SPO+ loss to be bounded on the hypothesis class and the true conditional expectation to lie inside the fixed-lag linear autoregressive model, and neither is assured when predictions are unrestricted or the underlying dynamics are nonlinear.

Editorial extensions

If this is right

Training autoregressive forecasters with SPO+ loss is justified for stationary dependent cost sequences, not only for i.i.d. samples.
For bounded polyhedral feasible sets, the excess SPO risk decays as $O(m^{-1/4})$, a slower rate than the i.i.d. case, quantifying the price of dependence.
For strongly convex feasible sets the rate improves to $O(m^{-1/2})$, matching the i.i.d. calibration rate in terms of block count.
The bounds degrade as the mixing coefficient $\beta(a-l)$ grows and cease to be informative for non-mixing or unstable systems, matching the paper's experimental finding that regret rises sharply near spectral radius one.
The proposed algorithm can be implemented by stochastic subgradient descent because a subgradient of the SPO+ surrogate is available in closed form.
In the knapsack experiments, SPO+ beats the absolute loss and least squares loss when cost vectors come from nonlinear stationary dynamics, especially as the nonlinearity degree increases.

Reading between the lines

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

A natural practical extension is to tune the block length $a$: longer blocks reduce the mixing penalty $2m\beta(a-l)$ but also cut the effective sample size $m \approx n/(2a)$, so the optimal block size trades off bias and variance.
The blocking-plus-calibration template is not SPO-specific; any surrogate loss with a calibration function and bounded loss difference should admit the same dependent-data treatment.
For real load-forecasting data, a direct test would estimate $\beta(k)$ from the residuals and compare the predicted confidence degradation $\delta' = \delta - 2m\beta(a-l)$ against empirical coverage.
The experiments suggest SPO+ retains its advantage on nonlinear dynamics even though the theory assumes a linear fixed-memory class; a fully matched theory would need a nonlinear function class with its own complexity measure.

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.

Desk Editor's Note

read the letter

The paper has a real idea: extend smart predict-then-optimize to dependent time series using an autoregressive model and blocking arguments. The SPO generalization bound (Theorem 1) is a standard application of Mohri-Rostamizadeh and Yu's blocking technique, and it is fine. The autoregressive formulation is sensible, and the experiments, while limited, do show SPO+ helping on a nonlinear knapsack instance.

But the central calibration results do not hold as stated. Theorem 2's proof applies Theorem 1 to the SPO+ loss via equation (29), and Theorem 1 is only proved for the SPO loss. That would require ℓ_SPO+ to be bounded, and it is not: with S=[0,1], y=1, and ŷ→-∞, ℓ_SPO+(ŷ,y) grows like 2|ŷ|. Since the autoregressive prediction is an unconstrained linear map, the predicted cost vectors are unbounded and the Rademacher complexity is infinite. The proof also silently swaps the stated bound on R_{SPO}(H) for a bound on R_{SPO+}(H). This is not a cosmetic gap; it breaks the m^{-1/4} calibration rate. The fix is likely a bounded hypothesis class or a truncation argument, but it needs to be stated and propagated through the proof. Theorem 3 inherits the same issue, plus a typo in the statement (polyhedron vs. strongly convex set).

A second concern is that the experiments use a nonlinear observer, so the assumption that the true conditional expectation lies in the fixed-memory linear class H is violated. That makes the experiments illustrative but not a test of the theory. No code or data are provided, and some experimental details are missing.

The reader's stress-test note is accurate: the boundedness gap is the load-bearing flaw. That said, the paper is not incoherent; it is a serious but incomplete attempt at a useful extension. With a boundedness assumption and a cleaned-up proof, the calibration results could plausibly be repaired.

For whom is this paper? Researchers working on decision-focused learning for time series. I would not cite it in its current form, but I would send it to a referee: the problem is relevant, and the gap is identifiable and potentially fixable. A desk reject would be too harsh; a major-revision recommendation is right.

Referee Report

4 major / 7 minor

Summary. The paper studies predict-then-optimize (PTO) methods when the training sequence is a stationary β-mixing process. It proposes a fixed-memory autoregressive model trained with the SPO+ surrogate loss, states a generalization bound for the SPO loss under dependent data (Theorem 1), and extends the uniform calibration results of Liu and Grigas between SPO+ and SPO losses to this dependent setting for polyhedral feasible sets (Theorem 2) and strongly convex level sets (Theorem 3). The paper also reports knapsack experiments comparing SPO+ with ℓ1 and ℓ2 losses on cost trajectories generated by a stochastic dynamical system. The main theoretical contribution is the claimed m^{-1/4} and m^{-1/2} calibration rates for the dependent autoregressive model.

Significance. If valid, the calibration bounds would be a useful extension of the i.i.d. results of Liu and Grigas to time-series data and would provide theoretical support for training autoregressive forecasters with the SPO+ loss. The paper's use of blocking to handle β-mixing is a sensible standard technique, and Theorem 1 is a plausible adaptation of existing Rademacher bounds for bounded losses. The experiments are reasonably extensive and show SPO+ can outperform standard regression losses. However, the central calibration theorems are not established: the proof applies a SPO-loss bound to the SPO+ loss without proving the required boundedness, and the SPO+ loss is in fact unbounded on the stated hypothesis class. The experiments also violate the linearity assumption used in the theory. As a result, the main contribution is currently unsupported.

major comments (4)

[Sec. 4.2, Eq. (29)] The proof applies Theorem 1, which was proved for the SPO loss, to the SPO+ loss without establishing the required uniform boundedness. For the autoregressive class defined in Section 3, predictions are unconstrained linear maps to R^d, and for S=[0,1], y=1, w*(y)=1, one has ℓ_SPO+(ŷ,1) = max_{w∈[0,1]}(1-2ŷ)w + 2ŷ - 1 = 2ŷ - 1 for ŷ ≥ 1/2, so sup_{ŷ∈R} ℓ_SPO+ = ∞. Therefore R^{m,Ď}_{SPO+}(H) is infinite, the concentration inequality behind Eq. (29) has no finite constants, and the m^{-1/4} bound in Theorem 2 is vacuous. A boundedness or projection assumption on H, together with a proof that ℓ_SPO+ is bounded on that class, must be supplied before Theorems 2 and 3 can be considered established.
[Sec. 4.2, Theorem 2] The theorem statement assumes R^{m,Ď}_{SPO}(H) ≤ C'/√m, but the proof requires R^{m,Ď}_{SPO+}(H) ≤ C'/√m (see Eq. (29) and the following sentence). These are different Rademacher complexities; the SPO+ complexity is not bounded by the SPO complexity in any obvious way, and no such bound is provided. The proof therefore does not follow from the stated assumptions.
[Sec. 4.2, Theorem 3] The theorem statement says 'Suppose that the feasible region S is a bounded polyhedron' even though Theorem 3 is supposed to cover strongly convex level sets. The statement also switches between \hat h^m_{SPO+} in the premise and \hat g^m_{SPO+} in the displayed bound, and the proof is deferred with 'similarly' without giving the strongly-convex-specific argument. These inconsistencies make the theorem's scope and validity unclear.
[Sec. 5, Eq. (16)] The experimental observer o(x_k,ξ_k)=((Hx_k)^{deg}+0.5·1)·ξ_k is nonlinear in the state for deg up to 8, so the conditional expectation E[y_{N+1}|y_{1:N}] is generally not a fixed-memory linear function of the last l observations. This violates the assumption, used in Theorems 2 and 3, that the optimal predictor f* (or h*) lies in the linear hypothesis class H. The experiments therefore do not validate the theoretical regime; at best they provide heuristic evidence for the practical value of SPO+.

minor comments (7)

[Sec. 2.3, Definition 3] The mixing condition is written as β(k) → 0 as k → 0; it should be k → ∞.
[Sec. 2.1, Eq. (3)] The constant term 'c^T w*(y)' is undefined; the standard SPO+ loss uses 'y^T w*(y)' in that place.
[Appendix A, proof of Corollary 1] The notation ℓ_SPO(Y_{n+1} - f(Y_{1:n})) is not meaningful; the loss arguments should be the predicted and realized cost vectors.
[Algorithm 1] The step 'Partial Autocorrelation Function is used to determine l' is not described; the paper should state how l is chosen or cite a specific procedure.
[Sec. 5] The parameter 'deg' is used in the observer equation but never defined; it should be introduced explicitly.
[Sec. 4.2, Theorems 2 and 3] The theorems define δ' = δ - 2mβ(a-l), which can be nonpositive for small δ; as in Theorem 1, the statements should require δ > 2mβ(a-l) so that the probability expressions are well defined.
[Figures 2 and 3] The submitted text around the figures contains garbled glyphs and the captions are partially unreadable; the figures and captions need to be regenerated.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central bounds are assembled from external non-iid generalization results and an external calibration theorem, not from the paper's own target claims.

full rationale

The paper's Theorem 1 is an application of standard non-iid Rademacher bounds (Yu [15]; McDonald-Shalizi-Schervish [21]; Mohri-Rostamizadeh [12]) to the SPO loss. Theorems 2 and 3 then combine Theorem 1 with calibration functions imported from Liu-Grigas [2]; those authors are not among the present paper's authors, and the calibration theorem is not re-derived from the paper's target rates. No parameter is fitted to the claimed rates, and no conclusion is used as an input to its own proof. The gap flagged in Eq. (29) - applying Theorem 1 to the SPO+ loss without proving ℓ_SPO+ boundedness, and silently using an SPO+ Rademacher assumption instead of the stated SPO one - is a serious correctness/assumption gap, but it is not circularity: the proof does not assume the m^{-1/4} calibration result it claims to establish. There is no load-bearing self-citation, no authors-imported uniqueness theorem, and no renaming of a known result. Therefore the circularity score is 0.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The theory introduces no fitted constants. The autoregressive lag l is selected from data in experiments; block length a is a proof parameter constrained by the mixing coefficient. The calibration claims rest on several unstated assumptions: the SPO+ loss is bounded (false in general), the optimal predictor is in the fixed-memory linear class, and conditional distributions are Gaussian. These assumptions, not fitted values, are what the central claim depends on.

free parameters (2)

Autoregressive lag length l = not reported; determined by partial autocorrelation function (Algorithm 1, line 2)
The order of the AR model is an input to the theory and is selected from data in the experiments. The bounds hold for any fixed l, but the experiments do not report the selected values across the 1000 trials.
Block length a (and number of blocks m) = not specified; constrained by δ > 2mβ(a-l)
In Theorem 1 and the calibration theorems, the sample is partitioned as n = 2am + l. The choice of a trades off the O(1/√m) term against the mixing bias 2mβ(a-l). The paper never states how a should be chosen in practice, making the bounds non-constructive.

assumptions (6)

domain assumption The observed sequence {y_i} is stationary and β-mixing with mixing coefficients β(a) → 0
Assumption 1 in Section 2.3 underlies all bounds and the blocking argument.
domain assumption The predictor is a fixed-memory autoregression f(y_{1:i-1}) = f(y_{i-l:i-1}) with fixed lag l
Section 3 defines the model; Corollary 1 and the empirical risk in Eq. (9) rely on this.
ad hoc to paper The optimal predictor f* (or h* = E[y_{N+1}|y_{1:N}]) lies in the hypothesis class H
Theorem 2 and 3 state this; for the nonlinear experimental dynamics it is not satisfied, and the paper does not discuss misspecification.
domain assumption The conditional distributions P(·|x) belong to the Gaussian class P_Gaussian
Theorems 2 and 3 restrict to P_Gaussian following [2]; the experiments use multiplicative uniform noise and polynomial observers, which are not Gaussian.
ad hoc to paper The SPO+ loss is bounded by the same constant ω_S(Y) used in the SPO loss bound
Appendix B Eq. (29) applies Theorem 1 to the SPO+ loss. Boundedness of ℓ_SPO+ is never established and is false for unconstrained predictions, e.g., S=[0,1], y=1, ŷ=-10 gives ℓ_SPO+ = 21.
ad hoc to paper The Rademacher complexity of H with respect to the SPO+ loss is O(1/√m)
The proof of Theorem 2 uses R_{m,~D}^{SPO+}(H) ≤ C'/√m, but the theorem statement only assumes the same for the SPO loss.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Smart Predict-then-Optimize Method with Dependent Data: Risk Bounds and Calibration of Autoregression." pith.science (2026). https://pith.science/paper/ABR4IBQ7

@misc{pith2026241112653,
  author       = {Pith},
  title        = {Pith review of: Smart Predict-then-Optimize Method with Dependent Data: Risk Bounds and Calibration of Autoregression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ABR4IBQ7}},
  note         = {Machine review of arXiv:2411.12653}
}

read the original abstract

The predict-then-optimize (PTO) framework is indispensable for addressing practical stochastic decision-making tasks. It consists of two crucial steps: initially predicting unknown parameters of an optimization model and subsequently solving the problem based on these predictions. Elmachtoub and Grigas [1] introduced the Smart Predict-then-Optimize (SPO) loss for the framework, which gauges the decision error arising from predicted parameters, and a convex surrogate, the SPO+ loss, which incorporates the underlying structure of the optimization model. The consistency of these different loss functions is guaranteed under the assumption of i.i.d. training data. Nevertheless, various types of data are often dependent, such as power load fluctuations over time. This dependent nature can lead to diminished model performance in testing or real-world applications. Motivated to make intelligent predictions for time series data, we present an autoregressive SPO method directly targeting the optimization problem at the decision stage in this paper, where the conditions of consistency are no longer met. Therefore, we first analyze the generalization bounds of the SPO loss within our autoregressive model. Subsequently, the uniform calibration results in Liu and Grigas [2] are extended in the proposed model. Finally, we conduct experiments to empirically demonstrate the effectiveness of the SPO+ surrogate compared to the absolute loss and the least squares loss, especially when the cost vectors are determined by stationary dynamical systems and demonstrate the relationship between normalized regret and mixing coefficients.

Figures

Figures reproduced from arXiv: 2411.12653 by the authors.

**Figure 1.** SPO-based Autoregression Model Comparing simply minimizing the autoregressive model’s prediction error, such as least squares error and absolute error, by employing methods like regression, we design a more effective method that is based on the estimated regret of the desired decision by adopting the SPO and SPO+ loss into the autoregression processes. Essentially, SPO is tailored to conduct prediction focusing on t… view at source ↗

**Figure 2.** Normalized test set SPO loss for the SPO+, least squ [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** Boxplot of the test set’s normalized SPO loss for th [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 25 canonical work pages

[1]

predict, then o ptimize

Adam N Elmachtoub and Paul Grigas. Smart “predict, then o ptimize”. Management Science, 68(1):9–26, 2022

work page 2022
[2]

Risk bounds and calibration f or a smart predict-then-optimize method

Heyuan Liu and Paul Grigas. Risk bounds and calibration f or a smart predict-then-optimize method. Advances in Neural Information Processing Systems , 34:22083–22094, 2021

work page 2021
[3]

Melding the data-decisions pipeline: Decision-focused learning for combinatorial optimizatio n

Bryan Wilder, Bistra Dilkina, and Milind Tambe. Melding the data-decisions pipeline: Decision-focused learning for combinatorial optimizatio n. In Proceedings of the AAAI Confer- ence on Artiﬁcial Intelligence , volume 33, pages 1658–1665, 2019

work page 2019
[4]

Differentiation of blackbox combinatorial solvers

Marin Vlastelica Pogan ˇci´c, Anselm Paulus, Vit Musil, Georg Martius, and Michal Rolin ek. Differentiation of blackbox combinatorial solvers. In International Conference on Learning Representations, 2019

work page 2019
[5]

Optnet: Differentiable o ptimization as a layer in neural networks

Brandon Amos and J Zico Kolter. Optnet: Differentiable o ptimization as a layer in neural networks. In International Conference on Machine Learning , pages 136–145. PMLR, 2017

work page 2017
[6]

Task-based end-to-end model learning in stochastic optimization

Priya Donti, Brandon Amos, and J Zico Kolter. Task-based end-to-end model learning in stochastic optimization. Advances in neural information processing systems , 30, 2017

2017
[7]

From predictive t o prescriptive analytics

Dimitris Bertsimas and Nathan Kallus. From predictive t o prescriptive analytics. Management Science, 66(3):1025–1044, 2020

work page 2020
[8]

Generalization bounds in the predict-then-optimize framework

Othman El Balghiti, Adam N Elmachtoub, Paul Grigas, and A mbuj Tewari. Generalization bounds in the predict-then-optimize framework. Advances in neural information processing systems, 32, 2019

work page 2019

Show all 30 references

[9]

Fast rates for contextual linear optimization

Yichun Hu, Nathan Kallus, and Xiaojie Mao. Fast rates for contextual linear optimization. Management Science, 68(6):4236–4245, 2022

2022
[10]

En- ergy forecasting: A review and outlook

Tao Hong, Pierre Pinson, Yi Wang, Rafał Weron, Dazhi Y an g, and Hamidreza Zareipour. En- ergy forecasting: A review and outlook. IEEE Open Access Journal of Power and Energy , 7:376–388, 2020

2020
[11]

Discrepancy-base d theory and algorithms for forecast- ing non-stationary time series

Vitaly Kuznetsov and Mehryar Mohri. Discrepancy-base d theory and algorithms for forecast- ing non-stationary time series. Annals of Mathematics and Artiﬁcial Intelligence , 88(4):367– 399, 2020

2020
[12]

Rademacher com plexity bounds for non-iid pro- cesses

Mehryar Mohri and Afshin Rostamizadeh. Rademacher com plexity bounds for non-iid pro- cesses. Advances in Neural Information Processing Systems , 21, 2008

2008
[13]

Theory and algorit hms for forecasting time series

Vitaly Kuznetsov and Mehryar Mohri. Theory and algorit hms for forecasting time series. arXiv preprint arXiv:1803.05814, 2018

2018 arXiv
[14]

Convergence and consistency of regularized boosting algorithms with stationary b-mixi ng observations

Aurélie C Lozano, Sanjeev Kulkarni, and Robert E Schapi re. Convergence and consistency of regularized boosting algorithms with stationary b-mixi ng observations. Advances in neural information processing systems, 18, 2005

2005
[15]

Rates of convergence for empirical processes of stationary mixing sequences

Bin Y u. Rates of convergence for empirical processes of stationary mixing sequences. The Annals of Probability, pages 94–116, 1994

1994
[16]

Stability boun ds for non-iid processes

Mehryar Mohri and Afshin Rostamizadeh. Stability boun ds for non-iid processes. Advances in Neural Information Processing Systems , 20, 2007

2007
[17]

Cope: Trafﬁc engineering in dynamic networks

Hao Wang, Haiyong Xie, Lili Qiu, Y ang Richard Y ang, Yin Z hang, and Albert Greenberg. Cope: Trafﬁc engineering in dynamic networks. In Proceedings of the 2006 conference on Applications, technologies, architectures, and protocol s for computer communications , pages 99–110, 2006

2006
[18]

Prioritized allocation of emergency respon- ders based on a continuous-time incident prediction model

A yan Mukhopadhyay and Y evgeniy V orobeychik. Prioritized allocation of emergency respon- ders based on a continuous-time incident prediction model. In International Conference on Autonomous Agents and MultiAgent Systems , 2017

2017
[19]

Risk guarantee s for end-to-end prediction and optimization processes

Nam Ho-Nguyen and Fatma Kılınç-Karzan. Risk guarantee s for end-to-end prediction and optimization processes. Management Science, 68(12):8680–8698, 2022

2022
[20]

Sur l’extension du théorème limite du calcul des probabilités aux sommes de quantités dépendantes

Serge Bernstein. Sur l’extension du théorème limite du calcul des probabilités aux sommes de quantités dépendantes. Mathematische Annalen, 97:1–59, 1927

1927
[21]

Nonparametric risk bounds for time-series forecasting

Daniel J McDonald, Cosma Rohilla Shalizi, and Mark Sche rvish. Nonparametric risk bounds for time-series forecasting. Journal of Machine Learning Research , 18(32):1–40, 2017. 11

2017
[22]

How to compare different loss function s and their risks

Ingo Steinwart. How to compare different loss function s and their risks. Constructive Approx- imation, 26(2):225–287, 2007

2007
[23]

Adam: A method for stocha stic optimization

Diederik P Kingma and Jimmy Ba. Adam: A method for stocha stic optimization. arXiv preprint arXiv:1412.6980, 2014

2014 arXiv
[24]

Pyepo: A pytorch-based end-t o-end predict-then-optimize library for linear and integer programming

Bo Tang and Elias B Khalil. Pyepo: A pytorch-based end-t o-end predict-then-optimize library for linear and integer programming. arXiv preprint arXiv:2206.14234 , 2022

2022 arXiv
[25]

Mixing properties of arma proces ses

Abdelkader Mokkadem. Mixing properties of arma proces ses. Stochastic processes and their applications, 29(2):309–315, 1988

1988
[26]

Learning without mixing: Towards a sharp analysis of linear system id entiﬁcation

Max Simchowitz, Horia Mania, Stephen Tu, Michael I Jord an, and Benjamin Recht. Learning without mixing: Towards a sharp analysis of linear system id entiﬁcation. In Conference On Learning Theory, pages 439–473. PMLR, 2018

2018
[27]

Sample complexity of kalman ﬁltering for unknown systems

Anastasios Tsiamis, Nikolai Matni, and George Pappas. Sample complexity of kalman ﬁltering for unknown systems. In Learning for Dynamics and Control , pages 435–444. PMLR, 2020

2020
[28]

System id entiﬁcation: A machine learning per- spective

Alessandro Chiuso and Gianluigi Pillonetto. System id entiﬁcation: A machine learning per- spective. Annual Review of Control, Robotics, and Autonomous Systems , 2:281–304, 2019

2019
[29]

Sharp rates in depen- dent learning theory: Avoiding sample size deﬂation for the square loss

Ingvar Ziemann, Stephen Tu, George J Pappas, and Nikola i Matni. Sharp rates in depen- dent learning theory: Avoiding sample size deﬂation for the square loss. arXiv preprint arXiv:2402.05928, 2024

2024 arXiv
[30]

Least squares regression with markovian data: Fundamental limits and alg orithms

Dheeraj Nagaraj, Xian Wu, Guy Bresler, Prateek Jain, an d Praneeth Netrapalli. Least squares regression with markovian data: Fundamental limits and alg orithms. Advances in neural infor- mation processing systems, 33:16666–16676, 2020. 12 A Redemancher Complexity and the Gener...

2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

predict, then o ptimize

Adam N Elmachtoub and Paul Grigas. Smart “predict, then o ptimize”. Management Science, 68(1):9–26, 2022

work page 2022

[2] [2]

Risk bounds and calibration f or a smart predict-then-optimize method

Heyuan Liu and Paul Grigas. Risk bounds and calibration f or a smart predict-then-optimize method. Advances in Neural Information Processing Systems , 34:22083–22094, 2021

work page 2021

[3] [3]

Melding the data-decisions pipeline: Decision-focused learning for combinatorial optimizatio n

Bryan Wilder, Bistra Dilkina, and Milind Tambe. Melding the data-decisions pipeline: Decision-focused learning for combinatorial optimizatio n. In Proceedings of the AAAI Confer- ence on Artiﬁcial Intelligence , volume 33, pages 1658–1665, 2019

work page 2019

[4] [4]

Differentiation of blackbox combinatorial solvers

Marin Vlastelica Pogan ˇci´c, Anselm Paulus, Vit Musil, Georg Martius, and Michal Rolin ek. Differentiation of blackbox combinatorial solvers. In International Conference on Learning Representations, 2019

work page 2019

[5] [5]

Optnet: Differentiable o ptimization as a layer in neural networks

Brandon Amos and J Zico Kolter. Optnet: Differentiable o ptimization as a layer in neural networks. In International Conference on Machine Learning , pages 136–145. PMLR, 2017

work page 2017

[6] [6]

Task-based end-to-end model learning in stochastic optimization

Priya Donti, Brandon Amos, and J Zico Kolter. Task-based end-to-end model learning in stochastic optimization. Advances in neural information processing systems , 30, 2017

2017

[7] [7]

From predictive t o prescriptive analytics

Dimitris Bertsimas and Nathan Kallus. From predictive t o prescriptive analytics. Management Science, 66(3):1025–1044, 2020

work page 2020

[8] [8]

Generalization bounds in the predict-then-optimize framework

Othman El Balghiti, Adam N Elmachtoub, Paul Grigas, and A mbuj Tewari. Generalization bounds in the predict-then-optimize framework. Advances in neural information processing systems, 32, 2019

work page 2019