Invited Speakers

Sustainable Operations Management: Review and Future Directions

L. Beril Toktay
Georgia Tech Scheller College of Business

This talk will review the evolution of sustainable operations management research over the last 20 years, providing a “big picture” overview of trends in the sustainable operations management research in Operations Management journals, along with an impact assessment both inside and outside the boundaries of operations management research. We will conclude by proposing a number of future research directions.

Bio: L. Beril Toktay is Professor of Operations Management, Brady Family Chairholder and ADVANCE Professor at the Georgia Tech Scheller College of Business. Her primary research area is sustainable operations and supply chain management, with an emphasis on circular economy models. Her articles have appeared in Management Science, Manufacturing & Service Operations Management (M&SOM), Operations Research, Production and Operations Management (POM) and Industrial Ecology. Her research has been funded by several National Science Foundation and other foundation grants, and has been recognized as recipient of the Management Science Best Paper Award and the M&SOM Responsible Research Award. She held leadership roles including Area Editor in Operations Research and MSOM Society President, and is currently INFORMS VP of Marketing, Communications and Outreach. She became a Distinguished Fellow of the MSOM Society in 2017 and received the MSOM Distinguished Service Award in 2018. At Georgia Tech, she is the founding Faculty Director of the Ray C. Anderson Center for Sustainable Business and the Executive Faculty Co-Director of the Office of Serve-Learn-Sustain. Dr. Toktay obtained a BS in Industrial Engineering and Mathematics from Boğaziçi University, an MS in Industrial Engineering from Purdue University, a PhD in Operations Research from MIT, and was formerly Associate Professor of Operations Management at INSEAD.

Personalized Dynamic Pricing with Machine Learning

Bora Keskin
Duke University

This talk will feature state-of-the-art models and methods in personalized dynamic pricing with demand learning. In this context, we consider a seller who can dynamically adjust the price of a product at the individual customer level, by utilizing information about customers' characteristics encoded as a d-dimensional feature vector. We assume a personalized demand model, parameters of which depend on s out of the d features. The seller initially does not know the relationship between the customer features and the product demand, but learns this through sales observations over a selling horizon of T periods. We prove that the seller’s expected regret, i.e., the revenue loss against a clairvoyant who knows the underlying demand relationship, is at least of order sT^{1/2} under any admissible policy. We then design a near-optimal pricing policy for a “semi-clairvoyant” seller (who knows which s of the d features are in the demand model) that achieves an expected regret of order sT^{1/2}log(T). We extend this policy to a more realistic setting where the seller does not know the true demand predictors, and show this policy has an expected regret of order sT^{1/2}[log(d)+log(T)], which is also near-optimal. Finally, we test our theory on simulated data and on a data set from an online auto loan company in the United States. On both data sets, our experimentation-based pricing policy is superior to intuitive and/or widely-practiced customized pricing methods such as myopic pricing and segment-then-optimize policies. Furthermore, our policy improves upon the loan company’s historical pricing decisions by 47% in expected revenue over a six-month period.

Link for the Paper: https://ssrn.com/abstract=2972985

Bio: Bora Keskin is an Assistant Professor in the Operations Management area at the Fuqua School of Business at Duke University. Bora received his B.S. in Industrial Engineering and Mathematics from Boğaziçi University in 2007, and his Ph.D. from the Graduate School of Business at Stanford University in 2012. Before joining the faculty at Duke University in 2015, he worked at McKinsey & Company as a consultant in banking and telecommunications industries, and at the University of Chicago as an Assistant Professor of Operations Management. Bora's main research studies management problems that involve decision making under uncertainty. In particular he is interested in stochastic models and their application to revenue management, dynamic pricing, statistical learning, machine learning, and product differentiation. Bora has published papers in leading research journals such as Management Science, Operations Research, Manufacturing and Service Operations Management, and Mathematics of Operations Research. In 2019, Bora was awarded the Lanchester Prize for the development of a novel paradigm for the modeling and analysis of online dynamic optimization problems that are subject to temporal uncertainty.

Performance Guarantees for Network Revenue Management via Approximate Dynamic Programming

Hüseyin Topaloğlu
Cornell ORIE

In network revenue management problems, we have access to a set of resources with limited capacities. Requests for products sequentially arrive over time. If we accept a request for a product, then we generate a certain revenue and consume the capacities of a combination of resources, both depending on the product that is requested. The goal is to find a policy to decide which product requests to accept so as to maximize the total expected revenue over a finite selling horizon. Network revenue management problems find applications in a variety of settings. When selling airline tickets, for example, resources take the form of capacities on the flight legs and products take the form of itineraries with connecting flights. When managing cloud computing systems, resources take the form of computing capacities of different types and products take the form of requests for computing capacity for different durations. One can formulate network revenue management problems as a dynamic program, but the state variable in such a dynamic program ends up being a high-dimensional vector keeping the remaining capacities of the resources; hence solving the dynamic program is computationally prohibitive. We present an approach to approximate the value functions in the dynamic program. The approximations to the value functions have tunable parameters. We show how to tune the parameters so that the policies induced by our value function approximations have performance guarantees. In particular, if each product uses at most L resources, then our policy is guaranteed to obtain at least 1/(1+L) fraction of the optimal total expected revenue. This performance guarantee is near-optimal in the sense that improving it by more than a factor of log L is provably hard. Numerical work indicates that our policies perform remarkably well when compared with efficiently-computable upper bounds on the optimal total expected revenues. This is joint work with Yuhang Ma from Uber, Paat Rusmevichientong from University of Southern California, and Mika Sumida from Cornell University.

Bio: Huseyin Topaloglu is a Professor at Cornell Tech and in the Operations Research and Information Engineering Department at Cornell University. He received his Ph.D. from Princeton University in Operations Research and Financial Engineering in 2001. He has been a member of the Cornell faculty since 2002. He holds a B.S. degree in Industrial Engineering from Boğaziçi University. His research interests are in pricing, retail operations, logistics and supply chain management. Currently, he develops technology to price products such as airline tickets and hotel rooms in response to dynamic demand, and to find right product assortments to display to consumers in online retail by taking advantage of their past purchase patterns. He was the recipient of INFORMS Revenue Management Section Prize in 2010.

Fair and Interpretable Decision Rules for Binary Classification

Oktay Gunluk
Cornell ORIE

We consider the problem of building Boolean rule sets in disjunctive normal form (DNF), an interpretable model for binary classification, subject to fairness constraints. We formulate an integer program that maximizes classification accuracy with explicit constraints on two different measures of classification parity: equality of opportunity, and equalized odds. Column generation framework, with a novel formulation, is used to efficiently search over exponentially many possible rules, eliminating the need for heuristic rule mining. Empirical results on three standard fair machine learning datasets show that our algorithm performs well compared to other state-of-the-art methods.
Joint work with Connor Lawless

Bio: Oktay Gunluk joined the School of Operations Research and Information Engineering faculty in January 2020. Before joining Cornell, he was the manager of the Mathematical Optimization and Algorithms group at IBM Research. He has also spent three years as a researcher in the Operations Research group in AT&T Labs. At both of these industrial labs, in addition to basic research in mathematical optimization, he has worked on various large-scale applied optimization projects for internal and external customers. His main research interests are related to theoretical and computational aspects of discrete optimization problems, mainly in the area of integer programing. In particular, his main body of work is in the area of cutting planes for mixed-integer sets. Some of his recent work focuses on developing integer programming-based approaches to classification problems in machine learning. He has B.S./M.S. degrees in Industrial Engineering from Boğaziçi University (1987/1989), and M.S./Ph.D. degrees in Operations Research) from Columbia University (1993/1995).

Modeling Infectious Diseases, Evaluating Intervention Strategies, and Resource Allocation

Pınar Keskinocak
Georgia Tech School of Industrial and Systems Engineering

Infectious diseases continue to pose a major threat for populations around the world, despite much progress in science and medicine over the past decades. Mathematical models can help our understanding of disease progression in individuals and spread/prevalence across the population. Pharmaceutical and non-pharmaceutical interventions play an important role in infectious disease control. When there are limited resources for interventions, e.g., for prevention or treatment, modeling can also help in resource allocation, e.g., identifying which combinations of interventions would be most effective for different geographic regions, subpopulations, or individuals. In this presentation, we will share examples of such models and insights about how they might inform decisions in practice.

Bio: Pınar Keskinocak is the William W. George Chair and Professor in the School of Industrial and Systems Engineering and the co-founder and Director of the Center for Health and Humanitarian Systems at Georgia Institute of Technology. Dr. Keskinocak’s research focuses on the applications of quantitative methods and analytics to have a positive impact in society, particularly in healthcare and humanitarian systems. Her recent work has addressed a broad range of topics such as infectious disease modeling, evaluating intervention strategies, and resource allocation; catch-up scheduling for vaccinations; decision-support for organ transplant; hospital operations management; and disaster preparedness and response. She has worked on projects with a variety of governmental and non-governmental organizations, and healthcare providers, including American Red Cross, CARE, Carter Center, CDC, Children’s Healthcare of Atlanta, Emory Healthcare, Grady Hospital, and Task Force for Global Health. Her work has been published in numerous peer-reviewed journals. Dr. Keskinocak is an INFORMS Fellow, served as the 2020 President of INFORMS, and has served in various other roles within the society over the years, including INFORMS Secretary, INFORMS Vice President for Membership and Professional Recognition, President of the Women on OR/MS Forum, President of the Public Sector OR Section, and Department Editor for Operations Research. She has also been an active member of other professional societies including IISE.