-
作者：Hu, Yichun; Kallus, Nathan; Mao, Xiaojie
作者单位：Cornell University; Tsinghua University
摘要：We study a nonparametric contextual bandit problem in which the expected reward functions belong to a Holder class with smoothness parameter beta. We showhowthis interpolates between two extremes that were previously studied in isolation: nondifferentiable bandits (beta at most 1), with which rate-optimal regret is achieved by running separate noncontextual bandits in different context regions, and parametric-response bandits (infinite beta), with which rate-optimal regret can be achieved with...
-
作者：Park, Chiwoo; Do Noh, Sang; Srivastava, Anuj
作者单位：State University System of Florida; Florida State University; Sungkyunkwan University (SKKU); State University System of Florida; Florida State University
摘要：The analysis of motion and time has become significant in operations research, especially for analyzing work performance in manufacturing and service operations in the development of lean manufacturing and smart factory. This paper develops a framework for data-driven analysis of work motions and studies their correlations to work speeds or execution rates, using data collected from modern motion sensors. Past efforts primarily relied on manual steps involving time-consuming stop-watching, vid...
-
作者：Shahmoradi, Zahed; Lee, Taewoo
作者单位：University of Houston System; University of Houston
摘要：Inverse linear programming (LP) has received increasing attention because of its potential to infer efficient optimization formulations that can closely replicate the behavior of a complex system. However, inversely inferred parameters and corresponding forward solutions from the existing inverse LP methods can be highly sensitive to noise, errors, and uncertainty in the input data, limiting their applicability in data-driven settings. We introduce the notion of inverse and forward stability i...
-
作者：Nguyen, Viet Anh; Kuhn, Daniel; Esfahani, Peyman Mohajerin
作者单位：Stanford University; Swiss Federal Institutes of Technology Domain; Ecole Polytechnique Federale de Lausanne; Delft University of Technology
摘要：We introduce a distributionally robust maximum likelihood estimation model with a Wasserstein ambiguity set to infer the inverse covariance matrix of a p-dimensional Gaussian random vector from n independent samples. The proposed model minimizes the worst case (maximum) of Stein's loss across all normal reference distributions within a prescribed Wasserstein distance from the normal distribution characterized by the sample mean and the sample covariance matrix. We prove that this estimation pr...
-
作者：Farahani, Mehdi H.; Dawande, Milind; Janakiraman, Ganesh
作者单位：University of Miami; University of Texas System; University of Texas Dallas
摘要：We study the problem of managing queues in online food-ordering services where customers, who place orders online and pick up at the store, are offered a common quote time, that is, the promised pick-up time minus the time the order is placed. The objective is to minimize the long-run average expected earliness and tardiness cost incurred by the customers. We introduce the family of static threshold policies for managing such queues. A static threshold policy is one that starts serving the fir...
-
作者：Kallus, Nathan; Uehara, Masatoshi
作者单位：Cornell University
摘要：Off-policy evaluation (OPE) in reinforcement learning is notoriously difficult in long- and infinite-horizon settings due to diminishing overlap between behavior and target policies. In this paper, we study the role of Markovian and time-invariant structure in efficient OPE. We first derive the efficiency bounds and efficient influence functions for OPE when one assumes each of these structures. This precisely characterizes the curse of horizon: in time-variant processes, OPE is only feasible ...
-
作者：Smith, Zachary J.; Bickel, J. Eric
作者单位：University of Texas System; University of Texas Austin
摘要：This paper establishes a new relationship between proper scoring rules and convex risk measures. Specifically, we demonstrate that the entropy function associated with any weighted scoring rule is equal to the maximum value of an optimization problem where an investor maximizes a concave certainty equivalent (the negation of a convex risk measure). Using this connection, we construct two classes of proper weighted scoring rules with associated entropy functions based on phi-divergences. These ...
-
作者：Awiszus, Kerstin; Capponi, Agostino; Weber, Stefan
作者单位：Leibniz University Hannover; Columbia University
摘要：We study the ex ante minimization of market inefficiency, defined in terms of minimum deviation of market prices from fundamental values, from a centralized planner's perspective. Prices are pressured from exogenous trading actions of leveragetargeting banks, which rebalance their portfolios in response to asset shocks. We characterize market inefficiency in terms of two key drivers, the banks' systemic significance and the statistical moments of asset shocks, and develop an explicit expressio...
-
作者：Lassance, Nathan; DeMiguel, Victor; Vrins, Frederic
作者单位：University of London; London Business School
摘要：A natural approach to enhance portfolio diversification is to rely on factor-risk parity, which yields the portfolio whose risk is equally spread among a set of uncorrelated factors. The standard choice is to take the variance as risk measure, and the principal components (PCs) of asset returns as factors. Although PCs are unique and useful for dimension reduction, they are an arbitrary choice: any rotation of the PCs results in uncorrelated factors. This is problematic becausewe demonstrate t...
-
作者：Bertsimas, Dimitris; Shtern, Shimrit; Sturt, Bradley
作者单位：Massachusetts Institute of Technology (MIT); Technion Israel Institute of Technology; University of Illinois System; University of Illinois Chicago; University of Illinois Chicago Hospital
摘要：We investigate a simple approximation scheme, based on overlapping linear decision rules, for solving data-driven two-stage distributionally robust optimization problems with the type-infinity Wasserstein ambiguity set. Our main result establishes that this approximation scheme is asymptotically optimal for two-stage stochastic linear optimization problems; that is, under mild assumptions, the optimal cost and optimal first-stage decisions obtained by approximating the robust optimization prob...
