Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Online Pricing and Allocation with Demand Learning and Fulfillment Cost

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

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

Pith's one-line read The paper claims that a lower-confidence-bound meta-algorithm over locally convex price intervals achieves $\widetilde O(\sqrt{T mn} + mn)$ regret for joint pricing, inventory, and fulfillment allocation, matching the…

desk verdict The joint pricing-inventory-allocation formulation is genuinely new, but the central convexity lemma (Lemma B.2) is false, so the claimed regret bound is unproven. read the letter →

arxiv 2501.18049 v3 pith:O5P74MJ2 submitted 2025-01-29 cs.LG math.OCstat.ML

Jianyu Xu , Xuan Wang , Yu-Xiang Wang , Jiashuo Jiang This is my paper · ORCID
classification cs.LGmath.OCstat.ML MSC 90C1590B80
keywords dynamicpricingonlineresourceallocationinventorycontrolconvexoptimizationlower-confidenceboundregretminimizationlineardemandpiecewiseobjectives
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

A seller must choose a uniform price and warehouse stock levels each period, then fulfill randomly realized demand through a transportation network with heterogeneous per-unit costs. The paper's central claim is that this joint (price, inventory) decision can be learned online with $\widetilde O(\sqrt{T mn} + mn)$ regret, optimal in $T$ up to logarithmic factors. If correct, this is the first near-optimal algorithm for online pricing combined with an allocation LP, and it shows that the globally non-convex, non-smooth objective can still be optimized by decomposing the price line into intervals of local convexity. The seller only needs the demand to be linear-and-noisy; the algorithm learns the unknown demand parameters from bandit price feedback and full-information inventory feedback.

What carries the argument

The key machinery is the optimistic cost function $W(p)=\min_{\vec I} Q(\vec I,p)$—the minimal expected net loss over inventory given price—together with the piecewise convexity Lemmas 3.3–3.4. On each interval $[C_{i_k,j_k}, C_{i_{k+1},j_{k+1}}]$ between sorted supply costs, $W(p)$ is convex, which allows local OCO agents to operate. Each agent $A_K$ runs a doubling sub-epoch scheme that estimates $W$ at three quarter-points, compares empirical values against a confidence width $\Delta_K$, and either narrows the search interval by $3/4$ (horizontal convergence) or updates the error bar (vertical convergence). The lower-confidence-bound rule $LCB_K = \hat W_K - \frac34 \Delta_K$ decides which agent to explore next, and the analysis enforces both $\Delta_K = \widetilde O(1/\sqrt{T_K})$ and per-agent sub-regret $\widetilde O(\sqrt{T_K})$.

What would settle it

Evaluate Lemma B.2 on the per-period cost with $f(x;\theta)=\theta x$ and $S(x)=[0,1-x]$: if the pointwise maximum $Q(x)=x(1-x)$—a concave function—arises as the actual second-stage value, the epigraph interchange in the lemma cannot be correct, and the interval-convexity premise fails. More generally, simulate the algorithm's regret on a two-warehouse, one-consumer instance with linear demand and compare the empirical regret trajectory to $\widetilde O(\sqrt T)$ to see whether the bound holds computationally.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the population cost $Q(\vec I, p)$—inventory cost plus expected optimal fulfillment loss—is marginally convex in price on each interval between consecutive sorted transportation costs $C_{i_k,j_k}$, even though it is globally non-convex and non-smooth in $p$. This piecewise structure lets the algorithm assign one zeroth-order online convex optimization agent to each of the $mn+1$ intervals and run them under a lower-confidence-bound meta-algorithm. The meta-algorithm selects the agent with the smallest estimated lower bound on its local optimum; each agent simultaneously shrinks its horizontal search interval and tightens its vertical value uncertainty. The regret decomposes into per-agent sub-regrets of order $\widetilde O(\sqrt{T_K})$ plus meta-level selection error, yielding the overall $\widetilde O(\sqrt{T mn} + mn)$ bound with probability $1-\epsilon$.

Load-bearing premise

The whole argument hangs on the claim that, for a linear demand model, the per-period expected cost is convex in the price inside each interval between two consecutive transportation costs, so that each local agent has a convex function to optimize.

Editorial extensions

If this is right

  • A seller can jointly set price and inventories in a multi-warehouse network without knowing the demand curve, losing only $\widetilde O(\sqrt{T mn} + mn)$ cumulative profit relative to the clairvoyant optimum.
  • The rate matches the information-theoretic lower bound for the single-supplier, single-consumer case, so the $T$-dependence cannot be improved in general.
  • The algorithm runs in polynomial time for rational-polytope inventory sets, pairing the theoretical guarantee with a computable procedure.
  • The proof technique—decomposing the non-convex price domain into locally convex intervals and orchestrating local OCO agents with an LCB meta-algorithm—provides a template for other non-convex learning-to-optimize problems with bandit feedback.
  • Under non-linear demand families the same decomposition would yield $\widetilde O(T^\alpha (mn)^{1-\alpha})$ regret, with $\alpha \ge 1/2$ depending on the demand family, interpolating between the current rate and slower continuum-bandit rates.

Reading between the lines

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

  • The horizontal/vertical confidence machinery is not specific to transportation LPs; the same two-sided shrinkage could apply to any bandit problem whose value function is piecewise convex with known breakpoints, suggesting testable extensions to assortment or network-revenue settings.
  • If the interval-convexity lemma were to fail for some demand model, the LCB/OCO decomposition would lose its convex objective; checking that lemma is the fastest falsification of the claimed rate.
  • The uniform-price assumption means the algorithm cannot exploit consumer heterogeneity; a natural extension is to allow per-consumer prices, which would add an $n$-dimensional price search and likely change the metric from $mn$ to a dependence on the price-action geometry.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper studies an online joint pricing, inventory, and allocation problem with heterogeneous supply costs and unknown linear demand. It proposes a hierarchical algorithm (OCSAA) that uses a lower-confidence-bound meta-strategy over multiple zeroth-order online convex optimization agents, one per interval between sorted supply costs. The main theoretical claim is that Algorithm 2 achieves O~(sqrt(T mn) + mn) regret with high probability, matching the lower bound in T from Broder and Rusmevichientong (2012). The entire upper-bound analysis rests on structural claims that the per-period cost Q_t(I,p) is marginally convex in p on each cost interval (Lemma 3.3) and that the optimistic cost W(p) is piecewise convex (Lemma 3.4); these are proved in Appendices B.2 and B.3.

Significance. If the result were valid, it would be a notable contribution: a near-optimal online algorithm for joint pricing, inventory, and allocation with heterogeneous fulfillment costs, a problem that is globally nonconvex and nonsmooth. The proposed LCB meta-algorithm over local OCO agents is an interesting design and the claimed regret rate is the right target. However, the central structural lemma used to establish piecewise convexity is false as stated, and the proof of that lemma is the load-bearing step for the regret guarantee. The upper-bound result is therefore not established, and the paper cannot currently be accepted.

major comments (3)
  1. [Appendix B.2, Lemma B.2 and Eq. (21)] Lemma B.2 is false as stated, and the proof's key step is the invalid epigraph interchange in Eq. (21). Take f(x;θ)=θx and S(x)=[0,1−x] for x∈[0,1]; all hypotheses of the lemma hold (f is convex and Lipschitz in x for each θ, and S(x) is a convex set that shrinks monotonically), but Q(x)=max_{θ∈S(x)} θx = x(1−x), which is concave on [0,1]. The equality asserted in Eq. (21), namely ⋃_x ⋂_{θ∈S(x)} epi f(·;θ) = ⋂_{θ∈S(0)} ⋃_{x∈S^{-1}(θ)} epi f(·;θ), does not hold; the left side is the epigraph of a concave function, while the right side is an intersection of convex epigraphs. Since Lemma B.2 is the only proof of Lemma 3.3, the marginal convexity of Q_t(·,p) on each cost interval is unproven.
  2. [Section 3.2, Lemmas 3.3–3.4 and Appendix B.3] The piecewise convexity of W(p) is load-bearing for the entire regret analysis: Lemma 5.2, Corollary B.8, Lemma 5.5, and Theorem 5.1 all use the convexity of W(p) on each interval [C_{i_k,j_k}, C_{i_{k+1},j_{k+1}}] to justify the quarter-point comparisons in Algorithm 3 and the LCB bounds. The proof of Lemma 3.4 in Appendix B.3 inherits Lemma 3.3 through Lemma B.3 and Lemma B.4, so it collapses as well. Without a valid proof of piecewise convexity, the agents in Algorithms 3–5 are not guaranteed a convex objective, and the O~(sqrt(T mn) + mn) regret bound of Theorem 5.1 is unsupported.
  3. [Appendix B.3, Lemmas B.3 and B.4] Even if Lemma 3.3 were repaired, the proof of Lemma 3.4 contains additional unjustified analytic steps: Eq. (22) differentiates through the argmin I*(p) without establishing uniqueness or differentiability; Eq. (26) assumes the second derivative ∂²Q_t(I*(p),p)/∂p² exists and is nonnegative; and Eq. (27) uses Moore-Osgood-type limit interchanges without stating the required uniform convergence. These issues are secondary to the failure of Lemma B.2, but they would also need to be addressed in any revision.
minor comments (4)
  1. [Title and abstract] The arXiv metadata title and abstract ('Online Pricing and Allocation with Demand Learning and Fulfillment Cost') do not match the full-text title ('Joint Pricing and Resource Allocation: An Optimal Online-Learning Approach'); please align these.
  2. [Section 3.1] The term 'marginal convexity' is used repeatedly but never defined; please define it explicitly, including the domain of the argument that is being varied.
  3. [Algorithms 2 and 3] The arithmetic in lines such as 'a_{K,1}=3L_{K,1}+U_{K,1}/4' is ambiguous; it should be written as (3L_{K,1}+U_{K,1})/4, and similarly for b and c. Also, in Algorithm 3, line 23, the update 'b_{K,τ+1}←U_{K,τ}' appears to be a typo; it should presumably update the interval endpoint U_{K,τ+1}.
  4. [Section 5, Eq. (9)] The regret decomposition uses the notations P*_{K_t} and P* that are not defined; these should be p*_{K_t} and p*, respectively, to match the earlier notation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: regret analysis is self-contained; the cited lower bound is external and the Lemma B.2 objection is a correctness issue, not a circularity.

full rationale

No circular step is present in the paper's claimed derivation chain. The regret guarantee in Theorem 5.1 is derived from the stated linear-demand model, the LP structure of the second-stage allocation, concentration inequalities, and OCO sub-regret bounds; none of these inputs already contain the regret bound. The optimistic cost W(p) is defined through the offline LP rather than fitted to data, and no fitted parameter is renamed as a prediction. The claimed optimality in T relies on the lower bound of Broder and Rusmevichientong (2012), an external source with no author overlap with the present paper, so it is not a self-citation chain. The self-citations in the related-work section (Xu and Wang 2021-2024, Baby et al. 2023, Xu et al. 2023) are not load-bearing for Theorem 5.1. The skeptical concern about Lemma B.2 is a correctness objection: if the epigraph interchange in Eq. (21) is invalid, then Lemma 3.3 and the convexity used by the OCO agents are unproven, but an invalid proof step is not equivalent to circularity. No equation defines a target quantity in terms of itself, and no 'prediction' reduces by construction to an input. Thus the honest circularity finding is 0.

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

The central claim rests on the linear-demand model, boundedness, full-information feedback on I, and the false Lemma B.2 used to prove piecewise convexity. No parameters are fitted to data.

assumptions (5)
  • domain assumption Demand is linear and noisy: D_t = a - b p + N_t, with E[N_t]=0 and i.i.d. over time (Assumption 3.2).
    The entire algorithm and analysis assume this semi-parametric demand model; if demand is non-linear, the claimed rates are not derived.
  • domain assumption Boundedness of gamma, a, b, I, p (Assumption 3.6).
    Provides the Lipschitz constants and concentration bounds used throughout.
  • domain assumption Full knowledge of supply costs gamma, C_{i,j}, and bound parameters (Assumption 3.7).
    The agent intervals are defined by the sorted C_{i,j}; without this, the decomposition is unavailable.
  • domain assumption Seller observes the full realized cost function Q_t(I,p) for all I after seeing demand (full-information feedback on I).
    Algorithms 2-5 record the marginal function of I, not just the realized scalar cost.
  • ad hoc to paper Lemma B.2: the maximum over a monotonically shrinking convex set of convex Lipschitz functions is convex.
    The paper's proof is invalid; the epigraph set-equality is false, and the statement itself is false in general. This is a load-bearing unproved assertion.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Online Pricing and Allocation with Demand Learning and Fulfillment Cost." pith.science (2026). https://pith.science/paper/O5P74MJ2

@misc{pith2026250118049,
  author       = {Pith},
  title        = {Pith review of: Online Pricing and Allocation with Demand Learning and Fulfillment Cost},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O5P74MJ2}},
  note         = {Machine review of arXiv:2501.18049}
}
abstract

We study online learning for a seller that jointly chooses per-period inventory positions and a uniform price, then fulfills realized demand through a downstream allocation. The main difficulty is not only demand learning: the price shifts demand and reshapes the transportation LP, making the population objective globally non-convex and non-smooth. To solve this problem, we propose OCSAA, an algorithm that exploits demand observations through counterfactual translation and proposes joint (price, inventory) decisions through lower-confidence optimism. OCSAA admits a polynomial-time additive-accuracy implementation for rational-polytope inventory sets. We prove a high-probability $\widetilde O(\sqrt T)$ regret guarantee and establish a matching-in-$T$ information-theoretic lower bound. Our results illustrate an effective integration of statistical learning methodologies with complex operations research problems.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Two-Timescale Hierarchical Reinforcement Learning for Resilient Operations

    stat.ML 2026-07 conditional novelty 6.0 of 10

    Synchronized two-timescale hierarchical PPO-style learning converges in average optimality gap at O(T^{-1/2}) (faster under market sharpness) and raises simulated used-car profits under joint shocks.

Reference graph

Works this paper leans on

49 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    P., Hsu, D

    Agarwal, A., Foster, D. P., Hsu, D. J., Kakade, S. M., and Rakhlin, A. (2011). Stochastic convex optimization with bandit feedback. Advances in Neural Information Processing Systems , 24

    work page 2011

  2. [2]

    Asadpour, A., Wang, X., and Zhang, J. (2020). Online resource allocation with limited flexibility. Management Science , 66(2):642--666

    work page 2020

  3. [3]

    Baby, D., Xu, J., and Wang, Y.-X. (2023). Non-stationary contextual pricing with safety constraints. Transactions on Machine Learning Research

    work page 2023

  4. [4]

    and Keskin, N

    Ban, G.-Y. and Keskin, N. B. (2021). Personalized dynamic pricing with machine learning: High-dimensional features and heterogeneous elasticity. Management Science , 67(9):5549--5568

    2021

  5. [5]

    and Zeevi, A

    Besbes, O. and Zeevi, A. (2009). Dynamic pricing without knowing the demand function: Risk bounds and near-optimal algorithms. Operations Research , 57(6):1407--1420

    work page 2009

  6. [6]

    and Rusmevichientong, P

    Broder, J. and Rusmevichientong, P. (2012). Dynamic pricing under a general parametric choice model. Operations Research , 60(4):965--980

    2012

  7. [7]

    and Wang, H

    Bumpensanti, P. and Wang, H. (2020). A re-solving heuristic with uniformly bounded loss for network revenue management. Management Science , 66(7):2993--3009

    work page 2020

  8. [8]

    Chen, B., Chao, X., and Ahn, H.-S. (2019). Coordinating pricing and inventory replenishment with nonparametric demand learning. Operations Research , 67(4):1035--1052

    work page 2019

Show all 49 references
  1. [9]

    Chen, B., Chao, X., and Shi, C. (2021a). Nonparametric learning algorithms for joint pricing and inventory control with lost sales and censored demand. Mathematics of Operations Research , 46(2):726--756

    2021

  2. [10]

    Chen, B., Chao, X., and Wang, Y. (2020). Data-based dynamic pricing and inventory control with censored demand and limited price changes. Operations Research , 68(5):1445--1456

    2020

  3. [11]

    Chen, B., Wang, Y., and Zhou, Y. (2023). Optimal policies for dynamic pricing and inventory control with nonparametric censored demands. Management Science

    2023

  4. [12]

    and Gallego, G

    Chen, N. and Gallego, G. (2021). Nonparametric pricing analytics with customer covariates. Operations Research

    2021

  5. [13]

    Chen, Q., Jasin, S., and Duenyas, I. (2021b). Joint learning and optimization of multi-product pricing with finite resource capacity and unknown demand parameters. Operations Research , 69(2):560--573

    2021

  6. [14]

    Chen, X., Zhang, X., and Zhou, Y. (2021c). Fairness-aware online price discrimination with nonparametric demand models. arXiv preprint arXiv:2111.08221

    2021 arXiv

  7. [15]

    C., and Oh, M.-h

    Choi, Y.-G., Kim, G.-S., Choi, Y., Cho, W., Paik, M. C., and Oh, M.-h. (2023). Semi-parametric contextual pricing algorithm using cox proportional hazards model. In International Conference on Machine Learning , pages 5771--5786. PMLR

    2023

  8. [16]

    C., Elmachtoub, A

    Cohen, M. C., Elmachtoub, A. N., and Lei, X. (2022). Price discrimination with fairness constraints. Management Science

    2022

  9. [17]

    C., Lobel, I., and Paes Leme, R

    Cohen, M. C., Lobel, I., and Paes Leme, R. (2020). Feature-based dynamic pricing. Management Science , 66(11):4921--4943

    2020

  10. [18]

    C., Miao, S., and Wang, Y

    Cohen, M. C., Miao, S., and Wang, Y. (2021). Dynamic pricing with fairness constraints. Available at SSRN 3930622

    2021

  11. [19]

    Cournot, A. A. (1897). Researches into the Mathematical Principles of the Theory of Wealth . Macmillan

  12. [20]

    Eyster, E., Madar \'a sz, K., and Michaillat, P. (2021). Pricing under fairness concerns. Journal of the European Economic Association , 19(3):1853--1898

    2021

  13. [21]

    Fan, J., Guo, Y., and Yu, M. (2021). Policy optimization using semiparametric models for dynamic pricing. arXiv preprint arXiv:2109.06368

    2021 arXiv

  14. [22]

    J., Simchi-Levi, D., and Wang, H

    Ferreira, K. J., Simchi-Levi, D., and Wang, H. (2018). Online network revenue management using thompson sampling. Operations research , 66(6):1586--1602

    2018

  15. [23]

    Hazan, E. (2016). Introduction to online convex optimization. Foundations and Trends in Optimization , 2(3-4):157--325

    2016

  16. [24]

    Hwang, D., Jaillet, P., and Manshadi, V. (2021). Online resource allocation under partially predictable demand. Operations Research , 69(3):895--915

    2021

  17. [25]

    and Kumar, S

    Jasin, S. and Kumar, S. (2012). A re-solving heuristic with bounded revenue loss for network revenue management with customer choice. Mathematics of Operations Research , 37(2):313--345

    2012

  18. [26]

    and Nazerzadeh, H

    Javanmard, A. and Nazerzadeh, H. (2019). Dynamic pricing in high-dimensions. The Journal of Machine Learning Research , 20(1):315--363

    2019

  19. [27]

    Jenatton, R., Huang, J., and Archambeau, C. (2016). Adaptive algorithms for online convex optimization with long-term constraints. In International Conference on Machine Learning , pages 402--411. PMLR

    2016

  20. [28]

    Jia, H., Shi, C., and Shen, S. (2022). Online learning and pricing with reusable resources: Linear bandits with sub-exponential rewards. In International Conference on Machine Learning , pages 10135--10160. PMLR

    2022

  21. [29]

    Jiang, J., Ma, W., and Zhang, J. (2022). Degeneracy is ok: Logarithmic regret for network revenue management with indiscrete distributions. arXiv preprint arXiv:2210.07996

    2022 arXiv

  22. [30]

    Karan, A., Balepur, N., and Sundaram, H. (2024). Designing fair systems for consumers to exploit personalized pricing. arXiv preprint arXiv:2409.02777

    2024 arXiv

  23. [31]

    B., Li, Y., and Song, J.-S

    Keskin, N. B., Li, Y., and Song, J.-S. (2022). Data-driven dynamic pricing and ordering with perishable inventory in a changing environment. Management Science , 68(3):1938--1958

    2022

  24. [32]

    Kleinberg, R. (2004). Nearly tight bounds for the continuum-armed bandit problem. Advances in Neural Information Processing Systems , 17:697--704

    2004

  25. [33]

    and Leighton, T

    Kleinberg, R. and Leighton, T. (2003). The value of knowing a demand curve: Bounds on regret for online posted-price auctions. In IEEE Symposium on Foundations of Computer Science (FOCS-03) , pages 594--605. IEEE

    2003

  26. [34]

    LaFrance, J. T. (1985). Linear demand functions in theory and practice. Journal of Economic theory , 37(1):147--166

    1985

  27. [35]

    P., Sivan, B., Teng, Y., and Worah, P

    Leme, R. P., Sivan, B., Teng, Y., and Worah, P. (2021). Learning to price against a moving target. In International Conference on Machine Learning , pages 6223--6232. PMLR

    2021

  28. [36]

    W., et al

    Luo, Y., Sun, W. W., et al. (2021). Distribution-free contextual dynamic pricing. arXiv preprint arXiv:2109.07340

    2021 arXiv

  29. [37]

    W., and Liu, Y

    Luo, Y., Sun, W. W., and Liu, Y. (2022). Contextual dynamic pricing with unknown noise: Explore-then-ucb strategy and improved regrets. In Advances in Neural Information Processing Systems

    2022

  30. [38]

    Reiman, M. I. and Wang, Q. (2008). An asymptotically optimal policy for a quantity-based network revenue management problem. Mathematics of Operations Research , 33(2):257--282

    2008

  31. [39]

    Shalev-Shwartz, S. et al. (2012). Online learning and online convex optimization. Foundations and Trends in Machine Learning , 4(2):107--194

    2012

  32. [40]

    and Wang, C

    Simchi-Levi, D. and Wang, C. (2023). Pricing experimental design: causal effect, expected revenue and tail risk. In International Conference on Machine Learning , pages 31788--31799. PMLR

    2023

  33. [41]

    Tullii, M., Gaucher, S., Merlis, N., and Perchet, V. (2024). Improved algorithms for contextual dynamic pricing. arXiv preprint arXiv:2406.11316

    2024 arXiv

  34. [42]

    and Banerjee, S

    Vera, A. and Banerjee, S. (2019). The bayesian prophet: A low-regret framework for online decision making. ACM SIGMETRICS Performance Evaluation Review , 47(1):81--82

    2019

  35. [43]

    Vera, A., Banerjee, S., and Gurvich, I. (2021). Online allocation and pricing: Constant regret via bellman inequalities. Operations Research , 69(3):821--840

    2021

  36. [44]

    Wang, H., Talluri, K., and Li, X. (2021a). On dynamic pricing with covariates. arXiv preprint arXiv:2112.13254

    2021 arXiv

  37. [45]

    Wang, Y., Chen, B., and Simchi-Levi, D. (2021b). Multimodal dynamic pricing. Management Science

    2021

  38. [46]

    Xu, J., Qiao, D., and Wang, Y.-X. (2023). Doubly fair dynamic pricing. In International Conference on Artificial Intelligence and Statistics , pages 9941--9975. PMLR

    2023

  39. [47]

    and Wang, Y.-X

    Xu, J. and Wang, Y.-X. (2021). Logarithmic regret in feature-based dynamic pricing. Advances in Neural Information Processing Systems , 34

    2021

  40. [48]

    and Wang, Y.-X

    Xu, J. and Wang, Y.-X. (2022). Towards agnostic feature-based dynamic pricing: Linear policies vs linear valuation with unknown noise. International Conference on Artificial Intelligence and Statistics (AISTATS)

    2022

  41. [49]

    and Wang, Y.-X

    Xu, J. and Wang, Y.-X. (2024). Pricing with contextual elasticity and heteroscedastic valuation. In Forty-first International Conference on Machine Learning

    2024

Pith tools

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