Présidents de session
Contextual stochastic programming: New frontiers for the sample average approximation in operations and statistics
- Bradley Sturt (University of Illinois Chicago)
Contextual stochastic programming
- Tito Homem-de-Mello (Universidad Adolfo Ibáñez)
Contextual stochastic programming: Contributed talks
- Jonathan Li (Telfer School of Management, University of Ottawa)
Optimal stopping is the problem of determining when to stop a stochastic system in order to maximize reward, which is of practical importance in domains such as finance, operations management and healthcare. Existing methods for high-dimensional optimal stopping that are popular in practice produce deterministic linear policies -- policies that deterministically stop based on the sign of a...
In this work, we explore a framework for contextual decision-making to study how the relevance and quantity of past data affects the performance of a data-driven policy. We analyze a contextual Newsvendor problem in which a decision-maker needs to trade-off between an underage and an overage cost in the face of uncertain demand. We consider a setting in which past demands observed under "close...
We consider a pricing problem in which the buyer is strategic: given the seller's pricing policy, the buyer can augment the features that they reveal to the seller in order to obtain a low price for the product. We model the seller's pricing problem as a stochastic program over an infinite-dimensional space of pricing policies in which the radii by which the buyer can strategically perturb...
The Application Driven Learning is a framework that integrates the predictive machine learning model training directly with the decision-making processes, optimizing predictions specifically for the application context.
We present ApplicationDrivenLearning.jl, a high-performance Julia package that enables efficient experimentation and implementation of the framework, particularly for...
We consider the class of two-stage stochastic programs with uncertainty only on the right-hand side. Such a class encompasses practical many problems, especially in inventory models. We show that, under certain conditions, there exist an optimal scenario, in the sense that solving the problem with that scenario yields the same optimal solution as the original problem. In the case data-driven...
We study contextual stochastic optimization problems in which the joint distribution of uncertain parameters and side information covariates is modeled as a mixture of Gaussians. In a data-driven setting, the parameters of this distribution are unknown and must be estimated from historical data. To mitigate the adverse effects of estimation errors and improve out-of-sample performance, we...
We consider data-driven decision-making that incorporates a prediction model within the 1-Wasserstein distributionally robust optimization (DRO) given joint observations of uncertain parameters and covariates using regression residuals in a streaming-data setting. In this setting, additional data becomes available and allows decisions to adapt to the growing knowledge of the underlying...
This paper proposes an Optimize-then-Predict framework in which we identify the optimal decision before predicting or observing the realized values. The optimization part can be run in low-demand environments, saving computational time during runtime. We also propose computationally efficient inferences for the evaluation of model performance.
This paper shows that in any optimization...
The recent interest in contextual optimization problems, where randomness is associated with side information, has led to two primary strategies for formulation and solution. The first, estimate-then-optimize, separates the estimation of the problem's parameters from the optimization process. The second, decision-focused optimization, integrates the optimization problem's structure directly...
We present a unified framework and practical solution for conditional risk minimization with side information—interpretable, tractable, and backed by finite-sample statistical guarantees.
This work addresses the design and planning of a two-echelon omnichannel distribution network under endogenous uncertainty. The network consists of suppliers, retail stores, and distribution centers (DC) that serve a dual role i.e., replenishing retail stores and fulfilling direct-to-customer deliveries. We focus on strategic decisions such as the opening and configuration of distribution...
We study learning in an adversarial setting, where an epsilon fraction of samples from a distribution P are globally corrupted (arbitrarily modified), and the remaining perturbations have an average magnitude bounded by rho (local corruptions). With access to n such corrupted samples, we aim to develop a computationally efficient approach that achieves the optimal minimax excess risk. Our...
