Distributionally Robust Performative Prediction

Songkai Xue; Yuekai Sun

REVIEW 3 major objections 4 minor 31 references

Distributionally Robust Performative Prediction

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper's core claim is that the distributionally robust performative optimum (DRPO) minimizes the worst-case performative risk over a KL-divergence ball of distribution maps, and when the true map falls inside that ball at the DRPO…

desk verdict Good new concept, clean duality, but the main excess-risk theorem (Prop 3.3) is false as stated: the proof drops the misspecification gap at the true PO, so the advertised localization to the true PO is not established. read the letter →

arxiv 2412.04346 v2 pith:7N3HBUVR submitted 2024-12-05 cs.LG stat.ML

Songkai Xue , Yuekai Sun This is my paper · ORCID

classification cs.LGstat.ML

keywords performativepredictiondistributionallyrobustoptimizationoptimumKLdivergencedistributionmapmisspecificationtiltedriskminimizationstrategicclassificationfairnesswithoutdemographics

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

Performative prediction wants a model that is good after it changes the data distribution it predicts, but the map describing that change is never known exactly. This paper proposes to optimize, instead of the nominal performative risk, the worst-case performative risk over a KL-divergence ball of distribution maps around the nominal map, and calls the minimizer the distributionally robust performative optimum (DRPO). The key claim is that when the true distribution map falls inside that ball at the DRPO, the DRPO's excess risk relative to the true optimum is bounded by the square root of the loss variance at the true optimum, a localized error that can be much smaller than the PO's error bound, which depends on the worst misspecification over the whole parameter space. This makes DRPO a robust approximation of the true PO under map misspecification, and the paper backs this with an efficient tilted-performative-risk reformulation and with experiments in strategic classification, partially identifiable maps, and fairness without demographics.

What carries the argument

The load-bearing object is the KL uncertainty collection $\mathcal{U}(D)$ around the nominal distribution map, with radius $\rho$ measured in KL divergence at each parameter $\theta$. The argument runs through the strong-duality identity $\operatorname{DRPR}(\theta)=\inf_{\mu\ge 0}\{ \mu \log \mathbb{E}_{Z\sim D(\theta)}[e^{\ell(Z;\theta)/\mu}] + \mu\rho \}$, which turns the infinite-dimensional supremum over maps into a one-dimensional dual problem and identifies the worst-case map as an exponential tilt of $D(\theta)$. This dual form lets the authors recast DRPO as the minimizer of a tilted performative risk $\mathbb{E}_{Z\sim D(\theta)}[e^{\alpha\ell(Z;\theta)}]$, and their excess-risk proof uses the cancellation $\operatorname{DRPR}(\theta_{\mathrm{DRPO}})-\operatorname{DRPR}(\theta_{\mathrm{PO,true}})\le 0$ plus a sensitivity expansion of the KL-DRO value at the true optimum to localize the error. The same decomposition gives a generalized bound with an extra infimum term when the coverage condition fails (Proposition C.1).

What would settle it

Take a one-dimensional location family with nominal map $D(\theta)=\mathcal{N}(\theta,\sigma^2)$, true map $D_{\mathrm{true}}(\theta)=\mathcal{N}((1+\delta)\theta,\sigma^2)$, quadratic loss $\ell(z;\theta)=(z-\theta)^2$, and vary $\rho$ and $\delta$. Compute the exact DRPO and true PO; if, for $\delta$ values that keep $D_{\mathrm{KL}}(D_{\mathrm{true}}(\theta_{\mathrm{DRPO}})\|D(\theta_{\mathrm{DRPO}}))\le \rho$, the ratio $E(\theta_{\mathrm{DRPO}})/\sqrt{\rho \operatorname{Var}_{D(\theta_{\mathrm{PO,true}})}[\ell]}$ does not stay bounded as $\rho\to 0$, Proposition 3.3 is false; if coverage is violated, compare the empirical excess risk against the two-term bound of Proposition C.1.

Watch

Extended reading notes

Core claim

Define the distributionally robust performative risk $\operatorname{DRPR}(\theta)=\sup_{\widetilde{D}\in\mathcal{U}(D)} \mathbb{E}_{Z\sim\widetilde{D}(\theta)}[\ell(Z;\theta)]$ with uncertainty collection $\mathcal{U}(D)=\{\widetilde{D}: D_{\mathrm{KL}}(\widetilde{D}(\theta)\|D(\theta))\le \rho \text{ for all }\theta\}$, where $D$ is the nominal distribution map and $\rho$ is the robustness radius. The distributionally robust performative optimum $\theta_{\mathrm{DRPO}}$ minimizes this quantity. The paper proves that if the true map is covered at the robust solution, $D_{\mathrm{KL}}(D_{\mathrm{true}}(\theta_{\mathrm{DRPO}})\|D(\theta_{\mathrm{DRPO}}))\le \rho$, then the excess risk over the true performative optimum satisfies $E(\theta_{\mathrm{DRPO}})\le \sqrt{\rho \operatorname{Var}_{Z\sim D(\theta_{\mathrm{PO,true}})}[\ell(Z;\theta_{\mathrm{PO,true}})]}+o(\sqrt{\rho})$, whereas the plain PO obeys $E(\theta_{\mathrm{PO}})\le \sqrt{2B}\sup_{\theta}\sqrt{D_{\mathrm{KL}}(D_{\mathrm{true}}(\theta)\|D(\theta))}$. The DRPO bound is local: it charges only the variance of the loss at the true optimum, while the PO bound is global and can be inflated by one badly misspecified $\theta$. The paper also shows DRPR minimization is equivalent to an augmented performative risk minimization, so off-the-shelf PO solvers can be used, and that in the Gaussian location example DRPO is exactly an $L^1$-regularized PO.

Load-bearing premise

The clean localized guarantee rests on the true distribution map being close to the nominal map at the robust optimum, in the sense that the true distribution induced at $\theta_{\mathrm{DRPO}}$ lies inside the KL ball of radius $\rho$ around the nominal distribution at $\theta_{\mathrm{DRPO}}$; if the true map falls outside that ball, the excess-risk bound gains an extra misspecification term and the localization weakens.

Editorial extensions

If this is right

Choosing $\rho$ to cover the true map makes the true performative risk of $\theta_{\mathrm{DRPO}}$ bounded by $\operatorname{DRPR}(\theta_{\mathrm{DRPO}})$, a certificate of out-of-sample performance that the ordinary PO does not offer.
Algorithmically, the dual identity reduces DRPO to alternating minimization in which the hard subproblem is exactly a performative risk minimization, so existing PO solvers can be reused with a tuned tilt parameter $\alpha$.
When misspecification is small, the DRPO bound and the PO bound are comparable, so robustness comes at no substantial cost; when misspecification grows, DRPO can outperform PO in worst-case performative risk, as shown for strategic classification and partially identified location models.
In the fairness experiment, increasing the tilt shrinks the performative-risk gap between majority and minority subpopulations while raising population risk, offering a demographic-free fairness lever.
All excess-risk statements carry over to general $\varphi$-divergences, so the same localized guarantee holds for other divergence-based uncertainty sets.

Reading between the lines

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

The localized form of the bound suggests a practical selection rule the authors only partially explore: if the loss variance at the true optimum is small, even a conservatively large $\rho$ leaves the DRPO close to the oracle, so calibrating $\rho$ against estimated variance rather than worst-case KL might be cheaper than their grid search.
Because the worst-case map in the KL ball is an exponential tilt, DRPO is implicitly regularizing the right tail of the performative loss; one testable corollary is that DRPO should improve tail quantiles such as the value-at-risk of the performative loss, not only its mean, in heavy-tailed settings.
The paper's Wasserstein-extension remark points to a concrete open target: replacing the variance term by a Lipschitz norm would let DRPO handle losses whose exponential moments diverge, which the KL dual requires.

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

Dear colleague,

You should know two things about this paper before deciding whether to trust it. First, the distributionally robust performative optimum (DRPO) is a genuinely new and sensible idea: minimize worst-case performative risk over a KL ball of distribution maps, and the dual reformulation as an augmented/tilted performative problem is clean and connects nicely to standard DRO machinery. Second, the paper's central theoretical guarantee — Proposition 3.3 — is not proved and appears false as stated.

The issue is in Appendix D.3. The proof bounds PRtrue(θDRPO) − PRtrue(θ*PO) by DRPR(θ*PO) − PRtrue(θ*PO), then asserts this equals √(ρ Var_{D(θ*PO)}[ℓ]) + o(√ρ) via the KL-DRO sensitivity theorem. That theorem controls DRPR(θ) − PR(θ), the gap to the nominal distribution. The additional term PR(θ*PO) − PRtrue(θ*PO) is not controlled by the assumption KL(Dtrue(θDRPO)∥D(θDRPO)) ≤ ρ; it can be O(1). The stress-test example is correct: a two-point model with D(A)=Bernoulli(0.1), Dtrue(A)=Bernoulli(0.11), D(B)=Bernoulli(0.1), Dtrue(B)=Bernoulli(0.9), and losses that make B the true PO while the DRPO is A, gives excess risk 0.45 versus a claimed bound of about 0.0095. Proposition C.1 inherits the same flaw at its θ*PO term.

The rest of the paper is in decent shape. The duality (Prop 3.1) is standard and correctly cited. The excess-risk bound for the PO (Prop 3.2) is a straightforward Pinsker argument. The experiments are honest and show that robustness helps under moderate-to-large misspecification, which is a useful qualitative observation even if the theory is currently overstating the guarantee. Algorithm 1's convergence claim is hand-wavy, though the appendix walks it back to case-by-case convergence. No code is provided, which is a minor reproducibility setback but not disqualifying.

What would fix this? Either strengthen the assumption to uniform coverage (KL(Dtrue(θ)∥D(θ)) ≤ ρ for all θ) — which would make the proof go through with an extra sup term — or add the misspecification gap at θ*PO to the bound, which weakens the localization claim substantially. As written, the central theorem is not salvageable in place.

This paper is worth a serious referee: the DRPO concept and the tilted reformulation are likely to be reused, and the empirical findings are suggestive. But the authors need to correct the theorem and its proof before the main claim can be trusted.

Best, [Your name]

Referee Report

3 major / 4 minor

Summary. The paper proposes a distributionally robust version of performative prediction: given a nominal distribution map D, it defines the distributionally robust performative risk DRPR(θ) as the worst-case expected loss over a KL ball of distribution maps around D, and studies the resulting solution concept DRPO. The main theoretical contribution is an excess-risk comparison claiming that, when the true map is covered by the KL ball at θDRPO, the DRPO has excess risk no larger than sqrt(ρ Var_{D(θPO,true)}[ℓ]) + o(sqrt ρ), thereby 'localizing' the error to the true performative optimum, whereas the standard PO has an excess-risk bound involving a supremum over the whole parameter space. The paper also gives a strong-duality reformulation, an alternating-minimization algorithm, a tilted-risk interpretation, calibration procedures for ρ, and experiments in strategic classification, partial identification, and fairness settings.

Significance. If the central theorem were correct, this would be a useful and timely contribution: it connects KL-divergence DRO with performative prediction, gives a clean dual reformulation, and provides a concrete robust alternative to the usual PO under distribution-map misspecification. The paper also honestly notes that its comparison of upper bounds is not tight, and the empirical study is informative. However, the central claim of Proposition 3.3 is false as stated, and the flaw propagates to Proposition C.1. Since the advertised 'localization to the true PO' is the paper's main theoretical selling point, the manuscript needs substantive revision before the claims can be accepted.

major comments (3)

[Proposition 3.3 and Appendix D.3] The proof of Proposition 3.3 is invalid at its last step. The paper claims DRPR(θ*PO) − PRtrue(θ*PO) = sqrt(ρ Var_{D(θ*PO)}[ℓ(Z; θ*PO)]) + o(sqrt ρ), citing the sensitivity property of KL-DRO. That sensitivity result controls DRPR(θ*PO) − PR(θ*PO), i.e., the gap with respect to the nominal distribution D(θ*PO), not with respect to the true distribution Dtrue(θ*PO). The remaining term PR(θ*PO) − PRtrue(θ*PO) is not controlled by the stated assumption D(Dtrue(θDRPO)∥D(θDRPO)) ≤ ρ. The statement is false as written: take Θ = {A,B}, Z = {0,1}, D(A) = Bernoulli(0.1), Dtrue(A) = Bernoulli(0.11), D(B) = Bernoulli(0.1), Dtrue(B) = Bernoulli(0.9), ℓ(1;A)=5, ℓ(0;A)=0, ℓ(0;B)=1, ℓ(1;B)=0, and ρ = 0.001. Then KL(Dtrue(A)∥D(A)) ≈ 0.00054 ≤ ρ, DRPR(A) ≈ 0.567, DRPR(B) ≈ 0.913, so θDRPO = A; the true PO is B, the excess risk is PRtrue(A) − PRtrue(B) = 0.45, while the right-hand side of (3.3) is about 0.0095. Coverage at θDRPO alone is therefore insufficient, and the theorem needs either coverage at θPO,true (or uniformly over Θ) or an added misspecification term.
[Proposition C.1 and Appendix D.4] Proposition C.1 inherits the same flaw. In the proof in Appendix D.4, the term DRPR(θ*PO) − PRtrue(θ*PO) is again replaced by the KL-DRO sensitivity bound, which only controls the gap to the nominal distribution D(θ*PO). The projection term involving P at θDRPO does not repair the unmodeled misspecification at θ*PO. The stated generalized bound is therefore not established, and the claim that it reduces to Proposition 3.3 when D(Dtrue(θDRPO)∥D(θDRPO)) ≤ ρ does not rescue either statement.
[Section 4.1 and Appendix I] The convergence claim for Algorithm 1 is unsupported. The text states that 'the alternating minimization algorithm in common practice guarantees global convergence (to stationary point)' and that strong convexity gives convergence to a global minimum, but no assumptions are stated under which ψ(θ, µ) is sufficiently well behaved, and θ-updates are only assumed to be solved by an unspecified performative-risk minimization method. Appendix I later says convergence guarantees must be established case-by-case, which is inconsistent with the main-text claim. Please provide a formal convergence theorem with explicit conditions or remove the unqualified claim.

minor comments (4)

[Throughout] There are numerous typos and small errors, including 'advatanges' in the abstract, 'Pinksker's' in Appendix D.2, 'hitogram' in Appendix B, 'Precicely' and 'estiamte' in Section 5.2, 'misspeicification' and 'indentified' in Section 5.2, 'Algirthm' in Appendix H, and 'reformualtion' in Appendix H.
[Section 5 and Appendix G.1] The main text says each shaded region shows the standard error of the mean from 30 trials, but Appendix G.1 says the training procedure is repeated for 10 trials. Please reconcile this inconsistency.
[Section 4.2] The sentence 'we should have θTPO with α equals θDRPO with ρ = (µ⋆)^{-1}(1/α)' is garbled. For fixed µ, argmin_θ ψ(θ, µ) equals argmin_θ E_{Z∼D(θ)}[e^{ℓ(Z;θ)/µ}], so the correspondence should be written explicitly as θTPO(α) = θDRPO(ρ(α)) with α = 1/µ⋆(ρ), making the dependence on ρ and α clear.
[Section 3.2] The sentence 'the excess risks of them are comparable, thus doing no harm' overclaims: comparing two upper bounds that are not tight does not establish that the actual excess risks are comparable, so this conclusion should be softened or removed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; Proposition 3.3 has a proof gap but it is not a feedback loop.

full rationale

The paper's derivation chain is self-contained. DRPO (Definition 2.5) is defined independently as the minimizer of DRPR (Definition 2.4), a KL-DRO risk around the nominal distribution map; this does not presuppose any result about the true PO. The main theoretical machinery is Proposition 3.1, whose proof cites the external KL-DRO duality [13], and Proposition 3.3, whose last step cites the external sensitivity theorem [9]. Neither citation is by the authors, and no load-bearing self-citation chain is present. The radius rho is a user-chosen hyperparameter; although Section 4.3 calibrates rho so that the coverage assumption D(Dtrue(theta_DRPO)||D(theta_DRPO)) <= rho holds, that only makes the conditional bound applicable, and the bound itself is not fitted to the experimental outcomes. The toy example and experiments are demonstrations, not definitions of the theoretical quantities. The one serious issue is Appendix D.3: the proof writes DRPR(theta*_PO) - PRtrue(theta*_PO) = sqrt(rho Var_{D(theta*_PO)}[ell]) + o(sqrt(rho)), attributing this to [9], but [9]'s sensitivity theorem controls DRPR(theta*_PO) - PR(theta*_PO), the gap to the nominal D(theta*_PO), not the gap to Dtrue(theta*_PO); the term PR(theta*_PO) - PRtrue(theta*_PO) is uncontrolled under the stated coverage assumption at theta_DRPO. That is a correctness gap in the proof as written, not a circular reduction: the claimed bound does not equal one of the paper's definitions or fitted quantities, and the missing term was not assumed into existence. It therefore does not raise the circularity score.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claims rest on standard DRO duality and sensitivity results, plus a domain assumption that the true map is within a KL ball around the nominal map at the DRPO. There are no invented physical or mathematical entities. The only free parameter is the radius ρ, which is a tuning parameter rather than a fitted constant in the theoretical derivations.

free parameters (1)

critical radius ρ = Calibrated in experiments (e.g., 0.01 to 0.04 in Section 5.1; via line search in Section 4.3)
The size of the KL uncertainty set is chosen by the practitioner or calibrated on data. It is not fitted to force the theoretical results, but it is a hand-chosen parameter that controls the behavior of the DRPO.

assumptions (5)

standard math KL divergence strong duality for DRO: DRPR(θ) = inf_μ≥0 { μ log E_{D(θ)}[e^{ℓ/μ}] + μρ }
Invoked in Proposition 3.1 and cited to Theorem 1 of [13]. This is an external mathematical result about KL-constrained optimization, not proved in the paper.
standard math Sensitivity of KL-DRO: excess risk scales as sqrt(ρ Var) + o(sqrt ρ)
Used to derive Proposition 3.3 and Proposition C.1, cited to [9]. This is a known sensitivity property of KL-divergence DRO.
domain assumption True distribution map lies inside the KL ball at the DRPO: D(Dtrue(θDRPO)||D(θDRPO)) ≤ ρ
This is the coverage assumption in Proposition 3.3 that makes the localized excess risk bound valid. If it fails, the bound gains an extra infimum term (Proposition C.1).
domain assumption Bounded loss function |ℓ(z;θ)| ≤ B for all z,θ
Assumed in Propositions 3.2 and C.1 to apply Pinsker's inequality and total variation bounds. Common in performative prediction analyses.
domain assumption The nominal distribution map D(·) is known and pointwise absolutely continuous with respect to the true map
The KL uncertainty set inherently requires absolute continuity of each candidate map with respect to D(θ). This is a modeling assumption about the family of plausible maps.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Distributionally Robust Performative Prediction." pith.science (2026). https://pith.science/paper/7N3HBUVR

@misc{pith2026241204346,
  author       = {Pith},
  title        = {Pith review of: Distributionally Robust Performative Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7N3HBUVR}},
  note         = {Machine review of arXiv:2412.04346}
}

read the original abstract

Performative prediction aims to model scenarios where predictive outcomes subsequently influence the very systems they target. The pursuit of a performative optimum (PO) -- minimizing performative risk -- is generally reliant on modeling of the distribution map, which characterizes how a deployed ML model alters the data distribution. Unfortunately, inevitable misspecification of the distribution map can lead to a poor approximation of the true PO. To address this issue, we introduce a novel framework of distributionally robust performative prediction and study a new solution concept termed as distributionally robust performative optimum (DRPO). We show provable guarantees for DRPO as a robust approximation to the true PO when the nominal distribution map is different from the actual one. Moreover, distributionally robust performative prediction can be reformulated as an augmented performative prediction problem, enabling efficient optimization. The experimental results demonstrate that DRPO offers potential advantages over traditional PO approach when the distribution map is misspecified at either micro- or macro-level.

Figures

Figures reproduced from arXiv: 2412.04346 by the authors.

**Figure 2.** Results of Experiment 5.2. Left: performative risk incurred by the PO and the TPO’s with [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** Results of Experiment 5.3. Performative risk of the population, the majority, and the minority, as the tilt α increases. The vertical band indicates the calibrated tilt αcal’s. Referencing to Example 2.3, we examine the scenario where the population distribution map is a mixture of two subpopulation distribution maps. We train a classifier using the population distribution map Dpop, but target at its performance o… view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Histogram of performative loss under Experiment 5.1 with [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]

**Figure 5.** Figure 5: Additional results of Experiment 5.1. Left: performative balanced error rate incurred [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]

**Figure 6.** Figure 6: Additional results of Experiment 5.2. Left: performative balanced error rate incurred by [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]

**Figure 7.** Figure 7: Additional results of Experiment 5.3. Performative accuracy of the population, the majority, [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 26 canonical work pages

[1]

Entropic value-at-risk: A new coherent risk measure

Amir Ahmadi-Javid. Entropic value-at-risk: A new coherent risk measure. Journal of Optimiza- tion Theory and Applications, 155:1105–1123, 2012

work page 2012
[2]

Data-driven stochastic programming using phi-divergences

Güzin Bayraksan and David K Love. Data-driven stochastic programming using phi-divergences. In The operations research revolution, pages 1–19. INFORMS, 2015

work page 2015
[3]

Sample out-of-sample inference based on wasserstein distance

Jose Blanchet and Yang Kang. Sample out-of-sample inference based on wasserstein distance. Operations Research, 69(3):985–1013, 2021

work page 2021
[4]

Quantifying distributional model risk via optimal transport

Jose Blanchet and Karthyek Murthy. Quantifying distributional model risk via optimal transport. Mathematics of Operations Research, 44(2):565–600, 2019

2019
[5]

Performative prediction in a stateful world

Gavin Brown, Shlomi Hod, and Iden Kalemaj. Performative prediction in a stateful world. In International Conference on Artificial Intelligence and Statistics, pages 6045–6061. PMLR, 2022

work page 2022
[6]

A robust learning approach for regression models based on distributionally robust optimization

Ruidi Chen and Ioannis C Paschalidis. A robust learning approach for regression models based on distributionally robust optimization. Journal of Machine Learning Research, 19(13), 2018

work page 2018
[7]

Approximate regions of attraction in learning with decision-dependent distributions

Roy Dong, Heling Zhang, and Lillian Ratliff. Approximate regions of attraction in learning with decision-dependent distributions. In International Conference on Artificial Intelligence and Statistics, pages 11172–11184. PMLR, 2023

work page 2023
[8]

Stochastic optimization with decision-dependent distribu- tions

Dmitriy Drusvyatskiy and Lin Xiao. Stochastic optimization with decision-dependent distribu- tions. Mathematics of Operations Research, 48(2):954–998, 2023

work page 2023

Show all 31 references

[9]

Statistics of robust optimization: A generalized empirical likelihood approach

John C Duchi, Peter W Glynn, and Hongseok Namkoong. Statistics of robust optimization: A generalized empirical likelihood approach. Mathematics of Operations Research, 2021

2021
[10]

Distributionally robust stochastic optimization with wasserstein distance

Rui Gao and Anton Kleywegt. Distributionally robust stochastic optimization with wasserstein distance. Mathematics of Operations Research, 48(2):603–655, 2023

2023
[11]

Robust empirical optimization is almost the same as mean–variance optimization

Jun-ya Gotoh, Michael Jong Kim, and Andrew EB Lim. Robust empirical optimization is almost the same as mean–variance optimization. Operations research letters, 46(4):448–452, 2018

2018
[12]

Near-optimal bayesian ambiguity sets for distributionally robust optimization

Vishal Gupta. Near-optimal bayesian ambiguity sets for distributionally robust optimization. Management Science, 65(9):4242–4260, 2019

2019
[13]

Kullback-leibler divergence constrained distributionally robust optimization

Zhaolin Hu and L Jeff Hong. Kullback-leibler divergence constrained distributionally robust optimization. Available at Optimization Online, 1(2):9, 2013

2013
[14]

How to learn when data reacts to your model: performative gradient descent

Zachary Izzo, Lexing Ying, and James Zou. How to learn when data reacts to your model: performative gradient descent. In International Conference on Machine Learning , pages 4641–4650. PMLR, 2021

2021
[15]

How to learn when data gradually reacts to your model

Zachary Izzo, James Zou, and Lexing Ying. How to learn when data gradually reacts to your model. In International Conference on Artificial Intelligence and Statistics, pages 3998–4035. PMLR, 2022

2022
[16]

Give me some credit

Kaggle. Give me some credit. https://www.kaggle.com/c/GiveMeSomeCredit/data, 2012

2012
[17]

Theoretical statistics: Topics for a core course

Robert W Keener. Theoretical statistics: Topics for a core course. Springer Science & Business Media, 2010. 10

2010
[18]

Making decisions under outcome performativity

Michael P Kim and Juan C Perdomo. Making decisions under outcome performativity. In 14th Innovations in Theoretical Computer Science Conference (ITCS 2023). Schloss-Dagstuhl- Leibniz Zentrum für Informatik, 2023

2023
[19]

The sample average approximation method for stochastic discrete optimization

Anton J Kleywegt, Alexander Shapiro, and Tito Homem-de Mello. The sample average approximation method for stochastic discrete optimization. SIAM Journal on optimization, 12 (2):479–502, 2002

2002
[20]

Recovering best statistical guarantees via the empirical divergence-based distribu- tionally robust optimization

Henry Lam. Recovering best statistical guarantees via the empirical divergence-based distribu- tionally robust optimization. Operations Research, 67(4):1090–1105, 2019

2019
[21]

State dependent performative prediction with stochastic approxima- tion

Qiang Li and Hoi-To Wai. State dependent performative prediction with stochastic approxima- tion. In International Conference on Artificial Intelligence and Statistics, pages 3164–3186. PMLR, 2022

2022
[22]

Tilted empirical risk minimization

Tian Li, Ahmad Beirami, Maziar Sanjabi, and Virginia Smith. Tilted empirical risk minimization. arXiv preprint arXiv:2007.01162, 2020

2007 arXiv
[23]

On tilted losses in machine learning: Theory and applications

Tian Li, Ahmad Beirami, Maziar Sanjabi, and Virginia Smith. On tilted losses in machine learning: Theory and applications. Journal of Machine Learning Research, 24(142):1–79, 2023

2023
[24]

Plug-in performative optimization

Licong Lin and Tijana Zrnic. Plug-in performative optimization. arXiv preprint arXiv:2305.18728, 2023

2023 arXiv
[25]

Stochastic opti- mization for performative prediction

Celestine Mendler-Dünner, Juan Perdomo, Tijana Zrnic, and Moritz Hardt. Stochastic opti- mization for performative prediction. Advances in Neural Information Processing Systems, 33: 4929–4939, 2020

2020
[26]

Outside the echo chamber: Optimizing the performative risk

John P Miller, Juan C Perdomo, and Tijana Zrnic. Outside the echo chamber: Optimizing the performative risk. In International Conference on Machine Learning, pages 7710–7720. PMLR, 2021

2021
[27]

Data-driven distributionally robust optimization using the wasserstein metric: performance guarantees and tractable reformulations

Peyman Mohajerin Esfahani and Daniel Kuhn. Data-driven distributionally robust optimization using the wasserstein metric: performance guarantees and tractable reformulations. Mathemati- cal Programming, 171(1-2):115–166, 2018

2018
[28]

Long term fairness for minority groups via performative distributionally robust optimization

Liam Peet-Pare, Nidhi Hegde, and Alona Fyshe. Long term fairness for minority groups via performative distributionally robust optimization. arXiv preprint arXiv:2207.05777, 2022

2022 arXiv
[29]

Performative prediction

Juan Perdomo, Tijana Zrnic, Celestine Mendler-Dünner, and Moritz Hardt. Performative prediction. In International Conference on Machine Learning, pages 7599–7609. PMLR, 2020

2020
[30]

Decision-dependent risk minimization in geometrically decaying dynamic environments

Mitas Ray, Lillian J Ratliff, Dmitriy Drusvyatskiy, and Maryam Fazel. Decision-dependent risk minimization in geometrically decaying dynamic environments. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 8081–8088, 2022

2022
[31]

Distributionally Robust Performative Prediction

Soroosh Shafieezadeh-Abadeh, Daniel Kuhn, and Peyman Mohajerin Esfahani. Regularization via mass transportation. Journal of Machine Learning Research, 20(103):1–68, 2019. Supplementary Materials for Distributionally Robust Performative Prediction This supplementary materials c...

2019

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

Entropic value-at-risk: A new coherent risk measure

Amir Ahmadi-Javid. Entropic value-at-risk: A new coherent risk measure. Journal of Optimiza- tion Theory and Applications, 155:1105–1123, 2012

work page 2012

[2] [2]

Data-driven stochastic programming using phi-divergences

Güzin Bayraksan and David K Love. Data-driven stochastic programming using phi-divergences. In The operations research revolution, pages 1–19. INFORMS, 2015

work page 2015

[3] [3]

Sample out-of-sample inference based on wasserstein distance

Jose Blanchet and Yang Kang. Sample out-of-sample inference based on wasserstein distance. Operations Research, 69(3):985–1013, 2021

work page 2021

[4] [4]

Quantifying distributional model risk via optimal transport

Jose Blanchet and Karthyek Murthy. Quantifying distributional model risk via optimal transport. Mathematics of Operations Research, 44(2):565–600, 2019

2019

[5] [5]

Performative prediction in a stateful world

Gavin Brown, Shlomi Hod, and Iden Kalemaj. Performative prediction in a stateful world. In International Conference on Artificial Intelligence and Statistics, pages 6045–6061. PMLR, 2022

work page 2022

[6] [6]

A robust learning approach for regression models based on distributionally robust optimization

Ruidi Chen and Ioannis C Paschalidis. A robust learning approach for regression models based on distributionally robust optimization. Journal of Machine Learning Research, 19(13), 2018

work page 2018

[7] [7]

Approximate regions of attraction in learning with decision-dependent distributions

Roy Dong, Heling Zhang, and Lillian Ratliff. Approximate regions of attraction in learning with decision-dependent distributions. In International Conference on Artificial Intelligence and Statistics, pages 11172–11184. PMLR, 2023

work page 2023

[8] [8]

Stochastic optimization with decision-dependent distribu- tions

Dmitriy Drusvyatskiy and Lin Xiao. Stochastic optimization with decision-dependent distribu- tions. Mathematics of Operations Research, 48(2):954–998, 2023

work page 2023