-
作者：Che, Ethan; Dong, Jing; Tong, Xin T.
作者单位：Columbia University; National University of Singapore
摘要：Stochastic gradient descent (SGD) is a powerful optimization technique that is particularly useful in online learning scenarios. Its convergence analysis is relatively well understood under the assumption that the data samples are independent and identically distributed (iid). However, applying SGD to policy optimization problems in operations research involves a distinct challenge: the policy changes the environment and thereby affects the data used to update the policy. The adaptively genera...
-
作者：Bu, Like; Dawande, Milind; Janakiraman, Ganesh
作者单位：University of Texas System; University of Texas Dallas
摘要：We study effective mechanisms for a manufacturer (buyer) to procure components of an assembly system under asymmetric information (including private costs and unobservable effort) and supply uncertainty. For each component, the buyer has access to an unreliable supplier whose production cost is private, input effort is unobservable, and production yield is uncertain. Further, both the buyer and the supplier have access to a more expensive but reliable supply source. The supplier also has acces...
-
作者：Chen, Xinyun; Hong, Guiyu; Liu, Yunan
作者单位：The Chinese University of Hong Kong, Shenzhen; Shanghai University of Finance & Economics; Amazon.com; North Carolina State University
摘要：We investigate an optimization problem in a queueing system where the service provider selects the optimal service fee p and service capacity & micro; to maximize the cumulative expected profit (the service revenue minus the capacity cost and delay penalty). The conventional predict-then-optimize (PTO) approach takes two steps: First, it estimates the model parameters (e.g., arrival rate and service-time distribution) from data; second, it optimizes a model taking these parameters as input. A ...
-
作者：Chen, Ningyuan; Li, Anran; Yang, Shuoguang
作者单位：University of Toronto; Chinese University of Hong Kong; Hong Kong University of Science & Technology
摘要：We consider the revenue maximization problem for an online retailer who plans to display in order a set of products differing in their prices and qualities. Consumers have attention spans, that is, the maximum number of products they are willing to view, and inspect the products sequentially before purchasing a product or leaving the platform empty-handed when the attention span gets exhausted. Our framework extends the wellknown cascade model in two directions: random attention spans of a rep...
-
作者：Podinovski, Victor V.; Papaioannou, Grammatoula
作者单位：Loughborough University
摘要：In data envelopment analysis, value judgements expressed as weight restrictions in multiplier models correspond to production tradeoffs in the dual envelopment models. Such tradeoffs are interpretable as simultaneous changes to the inputs and outputs that are assumed to be technologically possible for all decision-making units (DMUs) in the technology. The specification of production tradeoffs leads to the creation of additional DMUs, expansion of technology, and improved discriminating power ...
-
作者：Liu, Huikang; Wiesemann, Wolfram; Yue, Man-Chung
作者单位：Shanghai Jiao Tong University; Imperial College London; Hong Kong Polytechnic University
摘要：Factored Markov decision processes (MDPs) are a prominent paradigm within the artificial intelligence community for modeling and solving large-scale MDPs whose rewards and dynamics decompose into smaller, loosely interacting components. Through the use of value function approximations, dynamic Bayesian networks, and context-specific independence, factored MDPs can achieve an exponential reduction in the state space of an MDP and, thus, scale to problem sizes that are beyond the reach of classi...
-
作者：Shamsi, Davood; Luenberger, Robert; Ye, Yinyu
作者单位：Shanghai Jiao Tong University; Shanghai Institute for Mathematics & Interdisciplinary Sciences
摘要：This research note revisits the framework proposed in our earlier work and explores its conceptual and algorithmic connection to recent advances in dual-based online resource allocation-particularly the dual mirror descent method introduced previously. Both approaches address the challenge of making real-time sequential allocation decisions under dynamically revealed constraints. Although the dual mirror descent method relies on Bregman divergence to guide dual updates, our framework derives n...
-
作者：Bichuch, Maxim; Feinstein, Zachary
作者单位：State University of New York (SUNY) System; University at Albany, SUNY; University at Buffalo, SUNY; Stevens Institute of Technology
摘要：Within this work we consider an axiomatic framework for Automated Market Makers (AMMs). AMMs are smart contracts that set prices for swaps on a pool of assets. By imposing reasonable axioms on the underlying utility function, we are able to characterize the properties of the swap size of the assets and of the resulting pricing oracle. In providing these general axioms, we define a novel measure of price impacts that can be used to quantify those costs between different AMM constructions. We ha...
-
作者：Shapiro, Alexander; Zhou, Enlu; Lin, Yifan; Wang, Yuhao
作者单位：University System of Georgia; Georgia Institute of Technology
摘要：Stochastic optimal control with unknown randomness distributions has been studied for a long time, encompassing robust control, distributionally robust control, and adaptive control. We propose a new episodic Bayesian approach that incorporates Bayesian learning with optimal control. In each episode, the approach learns the randomness distribution with a Bayesian posterior and subsequently solves the corresponding Bayesian average estimate of the true problem. The resulting policy is exercised...
-
作者：Hartmann, Lorenz; Kauffeldt, T. Florian
作者单位：University of Basel
摘要：Suggestion for abstract without references: In this paper, we present the first axiomatic characterization of preferences that can be represented by a Choquet integral with respect to an exact capacity. The characterizing axiom, binary diversification, is novel and reflects an inclination for bets on events, thereby capturing a specific type of ambiguity aversion. Furthermore, we demonstrate that the three capacity classes balanced, exact, and convex fully exhaust all levels of our family of k...
