Please join us on Sunday, October 15th (4-5:15pm - CC West Room 211B) for the 2023 Dantzig Dissertation Session.
Introducing our finalists:
Xiao Lei, Columbia University
Title: Revenue Management in Video Games and with Fairness
Abstract:
Video games are the largest and fastest-growing segment of the entertainment industry, yet have received limited attention from the operations community. This thesis explores revenue management and matchmaking problems in video games, focusing on loot boxes, player engagement, and fairness issues. We consider optimal pricing and design of loot boxes, aiming to maximize revenue while providing customer protection. We also explore managing player engagement through optimal matchmaking policies and AI bots, using optimization and real data. Finally, we address the inequality induced by price discrimination in e-commerce platforms, considering fairness regulations and their impact on social welfare.
Paul Goelz, Carnegie Mellon University
Title: Social choice for social good: Proposals for democratic innovation from computer science
Abstract:
Driven by shortcomings of current democratic systems, practitioners and political scientists are exploring democratic innovations, i.e., institutions for decision-making that more directly involve constituents. In this thesis, we support this exploration via three approaches: we design practical algorithms for use in democratic innovations, we mathematically analyze the fairness properties of proposed decision-making processes, and we identify extensions of such processes that satisfy desirable properties.
Antonio Castellanos, Technion – Israel Institution of Technology
Title: Uncertainty in Service Systems: Performance Measure Estimation and Optimization Methods for Contact Centers with Information Uncertainty
Abstract:
In the quest to improve services, companies offer customers the opportunity to interact with their agents using texting. This has become a favorite channel of communication for customers with companies. However, text-based contact centers face operational challenges. Via data analysis of two North American contact centers, we find that the usual ways of analyzing and measuring quality in call centers give biased estimations of performance levels. The reason for this is the way customers and employees behave in these systems, which creates various types of information uncertainty. In this thesis we identify the main sources of such uncertainty and develop methods to cope with them. We do this by combining statistical and optimization methods.
Hannah Li, Stanford University
Title: Experimental Design and Decision-Making in Marketplace Platforms
Abstract:
Online platforms rely on experiments to aid decision-making. However, in marketplace platforms, prior work shows that treatment effect estimates can be biased due to interference, or users interacting with each other. This dissertation develops a structural modeling method to study interference in marketplace experiments. The work finds two main results that are helpful for practitioners: (1) Supply and demand imbalance balance is a primary factor determining which side of the market (supply or demand) should be randomized to minimize bias. (2) The work introduces a novel experimental design using Two-Sided Randomization and associated estimators that achieve relatively low bias in a wide range of supply and demand imbalance regimes.
Sean Sinclair, Cornell University
Title: Adaptivity, Structure, and Objectives in Sequential Decision-Making
Abstract:
We will consider designing methods for sequential decision-making (bandits, reinforcement learning) that leverage auxiliary data sources (imitation learning, exogenous datasets, geometric assumptions) based on the themes of adaptivity, structures, and objectives. We will specialize this framework in areas including memory management, fair resource allocation, and cloud computing. Central to this, we will additionally discuss our open-source code instrumentation and methodology to analyze the multi-criteria performance of algorithms on these problems.
------------------------------
Daniela Saban
Associate Professor
Stanford University
Palo Alto CA
------------------------------