Events

"Staffing and Pricing in Co-Sourced Call Centers"

Presentation by Professor Jeffrey P. Kharoufeh
Department of Industrial Engineering
Clemson University
 

Wednesday, September 25
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Increasingly, service organizations are electing to co-source some of their customer support functions, especially those handled by call centers. That is, rather than servicing requests exclusively with in-house agents, a portion of service capacity can be delegated to an external service provider. Indeed, the business of co-sourcing customer service centers has rapidly grown into a multinational, multibillion dollar industry. However, organizations must weigh the benefits of co-sourcing against the potential costs of surrendering control of their primary source of direct customer support. In this talk, Kharoufeh will present a joint queueing and stochastic programming framework to help call center managers decide how much of their demand should be co-sourced, and how much should be serviced in-house, when faced with uncertain and dynamic call arrival rates. Before the arrival rate is realized, the call center responds to the external service provider's prices by setting its in-house staffing levels and choosing the number of co-sourced agents to place on call in order to minimize its expected total staffing costs. Once the arrival rates are realized, the call center activates the on-call agents to ensure that quality-of-service (QoS) requirements are met. For its part, the external provider seeks to maximize expected total revenues over a finite contract period by setting per-agent holding and activation prices. Kharoufeh will model the interplay between the call center and external service provider as a leader-follower game (formulated as a bilevel program) in which the call center (the follower) solves a two-stage stochastic integer program with recourse. He will show that the bilevel program can be reformulated as a quadratically-constrained linear program to obtain the contractor's optimal per-agent holding and activation prices, and that the optimal staffing problem yields a highly-tractable, closed-form solution for a common set of QoS constraints. A numerical study demonstrates that cost reductions can be achieved over other common staffing rules in the presence of imperfect and asymmetric information.

Bio:

Jeffrey P. Kharoufeh is Professor and Chair of the Department of Industrial Engineering at Clemson University. His methodological areas of expertise are applied probability and stochastic modeling with applications in energy systems, queueing, reliability and maintenance optimization. His research has been funded by the National Science Foundation, the Air Force Office of Scientific Research, the National Reconnaissance Office, the Department of Veterans Affairs and other federal agencies. Dr. Kharoufeh earned a Ph.D. in Industrial Engineering and Operations Research from the Pennsylvania State University, where he was an inaugural Weiss Graduate Fellow. He currently serves as Associate Editor of Operations Research, Area Editor of Operations Research Letters and as a member of the editorial board of Probability in the Engineering and Informational Sciences. He is a Fellow of IISE and a professional member of INFORMS and the Applied Probability Society.

"The Power of Two Samples in Generative Adversarial Networks"

Presentation by Professor Sewoong Oh
Paul G. Allen School of Computer Science and Engineering
University of Washington
 

Wednesday, September 18
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Prof. Oh brings the tools from Blackwell’s seminal result on comparing two stochastic experiments from 1953, to shine a new light on a modern application of great interest: Generative Adversarial Networks (GAN). Binary hypothesis testing is at the center of training GANs, where a trained neural network (called a critic) determines whether a given samples from the real data or the generated (fake) data. By jointly training the generator and the critic, the hope is that eventually the trained generator will generate realistic samples. One of the major challenges in GAN is known as “mode collapse”; the lack of diversity in the samples generated by thus trained generators. Oh proposes a new training framework, where the critic is fed with multiple samples jointly (which Oh calls packing), as opposed to each samples separately as done in standard GAN training. With this simple but fundamental departure from standard GANs, experimental results show that the diversity of the generated samples improve significantly. Oh analyzes this practical gain by first providing a formal mathematical definition of mode collapse and making a fundamental connection between the idea of packing and the intensity of mode collapse. Precisely, Oh shows that the packed critic naturally penalizes mode collapse, thus encouraging generators with less mode collapse. The analyses critically rely on operational interpretation of hypothesis testing and corresponding data processing inequalities, which lead to sharp analyses with simple proofs. For this talk, Oh will assume no prior background on GANs. 

Bio:

Sewoong Oh is an Associate Professor of Paul G. Allen School of Computer Science and Engineering at University of Washington. He received his PhD from the department of Electrical Engineering at Stanford University. Following his PhD, he worked as a postdoctoral researcher at Laboratory for Information and Decision Systems (LIDS) at MIT. His research interest is in theoretical machine learning, including generative adversarial networks and saddle-point problems, and privacy and blockchains. He was co-awarded the best paper award at SIGMETRICS in 2015, NSF CAREER award in 2016 and GOOGLE Faculty Research Award.

"Calibration of Robust Empirical Optimization Problems"

Presentation by Professor Andrew Lim
Department of Analytics and Operations/Department of Finance
National University of Singapore
 

Wednesday, September 4
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Lim will discuss recent results on the out-of-sample properties of robust empirical optimization and develop a theory for data-driven calibration of the “robustness parameter” for worst-case maximization problems with concave reward functions. Building on the intuition that robust optimization reduces the sensitivity to model misspecification by controlling the spread of the reward distribution, Lim will show that the first-order benefit of a “little bit of robustness” is a significant reduction in the variance of the out-of-sample reward while the corresponding impact on the mean is almost an order of magnitude smaller. One implication is that a substantial reduction in the variance of the out-of-sample reward (i.e., sensitivity of the expected reward to model misspecification) is possible at little cost if the robustness parameter is properly calibrated. To this end, Lim will introduce the notion of a robust mean-variance frontier to select the robustness parameter and show that it can be approximated using resampling methods like the bootstrap. Examples show that robust solutions resulting from “open loop” calibration methods (e.g., selecting a 90% confidence level regardless of the data and objective function) can be very conservative out-of-sample, while selecting an ambiguity parameter that optimizes an estimate of the out-of-sample expected reward (e.g., via the bootstrap) with no regard for the variance is often insufficiently robust. Lim will also explain why the out-of-sample expected reward generated by the solution of a worst-case problem can sometimes exceed that of a sample-average optimizer.

Bio:

Andrew Lim is a Professor in the Department of Analytics and Operations and the Department of Finance at the National University of Singapore. Prior to that, he was a faculty member in the Department of Industrial Engineering and Operations Research at the University of California (Berkeley). He is a past recipient of an NSF CAREER Award and has served on the editorial boards of a number of journals including Operations Research, Management Science, and the IEEE Transactions on Automatic Control. He has a PhD from the Australian National University. His research interests are in the areas of stochastic control and optimization, decision making under uncertainty, robust optimization, and financial engineering.

"Quantitative Models for Decision-Support in Healthcare Applications"

Presentation by Professor Pinar Keskinoack
School of Industrial and Systems Engineering
Georgia Institute of Technology
 

Wednesday, May 1
3:15pm - Refreshments, Keller Hall 3-176
3:30pm - Graduate Seminar, Keller Hall 3-180

About:

With the goal of improving patient outcomes, efficiency, and effectiveness, quantitative models are increasingly used for decision-support in healthcare. In this presentation we will discuss a few applications from organ transplant, vaccination, screening, and workforce allocation decisions.

Bio:

Pinar Keskinoack is the William W. George Chair and Professor in the School of Industrial and Systems Engineering, and co-founder and Director of the Center for Health and Humanitarian Systems at Georgia Tech, she also serves as the College of Engineering ADVANCE Professor. Her research focuses on the applications of quantitative methods to have a positive impact in society, particularly in Center, CDC, Children’s Healthcare of Atlanta, Emory Healthcare, Grady Hospital, and Task Force for Global Health.

"Pricing Models for Online Food Ordering Platforms: Commission Rates and Delivery Fees"

Presentation by Professor Xuanming Su
The Wharton School
University of Pennsylvania
 

Wednesday, April 24
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Online food ordering platforms connect restaurants with customers and deliver food to their door.  In return, the platform charges restaurants a commission rate (fraction of sales) and charges customers a delivery fee.  These pricing parameters fluctuate widely in practice.  Some restaurants opt to pay a higher commission rate for enhanced exposure on the platform’s website.  Furthermore, delivery fees often vary based on distance and locations of the restaurant and the customer.  We develop stylized game-theoretic models to highlight best practices for the platform and partner restaurants.  Using a dataset from a medium-sized food ordering platform, we conduct empirical analyses of current practice and provide computational estimates of potential improvements.  This is joint work with Jaelynn Oh and Chloe Glaeser.

Bio:

Xuanming Su is the Murrel J. Ades Professor in Operations, Information, and Decisions Department at The Wharton School of the University of Pennsylvania.  He received his PhD from Stanford Graduate School of Business.  His main research area is operations management, with a focus on strategic customer behavior in the retail sector.

"Exploiting Decomposable Structure to Design Better Algorithms for Solving Integer Problems"

Presentation by Professor Martin Savelsbergh
H. Milton Stewart School of Industrial and Systems Engineering
Georgia Institute of Technology
 

Wednesday, April 17
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Optimization problems in which some or all of the variables are constrained to take integer values are of broad applicability in a wide range of fields, ranging from medicine and healthcare to banking and finance to environmental management and conservation. Over recent decades, exact algorithms for their solution have become faster and more efficient, culminating in a variety of commercial software platforms and public domain codes that provide exceptional capability for solving practical problems to optimality. However, this seems to have only increased the appetite of practitioners to solve ever-larger problems, which challenge the state-of-the-art. In this talk, we bring together two apparently disparate observations: (i) many practical problems have decomposable structure and (ii) despite the enormous strides in solution algorithms, one key element common to all of them, namely, the branching rule, has remained largely untouched since it was first presented in the 1960’s. Yet the branching rule defines how the search space is divided in the ``divide-and-conquer’’ paradigm that forms the basis of all exact algorithms; it is central to the algorithm. Here, we will describe a new idea for exploiting decomposable structure in problems to derive alternative, powerful, new branching rules. These rules are demonstrated to speed up commercial solvers by orders of magnitude, on two classes of problems having different characteristics. The potential to generalize these ideas will also be discussed.

Bio:

Martin Savelsbergh is a logistics and optimization specialist with over 25 years of experience in mathematical modeling, operations research, optimization methods, algorithm design, performance analysis, transport, supply chain management, and production planning. He has published over 160 research papers in many of the top operations research and optimization journals and has supervised more than 30 Ph.D. students. Martin has a track record of creating innovative techniques for solving large-scale optimization problems in a variety of areas, ranging from service network design, to last-mile and crowdsourced delivery, to ridesharing.  He has demonstrated an ability to design and implement highly sophisticated and effective optimization algorithms as well as an ability to analyze practical decision problems and translate the insights obtained into optimal business solutions.  Martin holds the James C. Edenfield Chair in the H. Milton Stewart School of Industrial and Systems Engineering (ISyE) at Georgia Institute of Technology. He is co-director of The Supply Chain and Logistics Institute (SCL). SCL coordinates all supply chain and logistics activities on the Georgia Tech campus.  Martin Savelsbergh is Editor-in-Chief of Transportation Science, one of the most prestigious academic journals in the area of transportation science and logistics.

"A Process Systems Engineering Perspective on Adjustable Robust Optimization"

Presentation by Professor Qi Zhang
Department of Chemical Engineering and Materials Science
University of Minnesota—Twin Cities
 

Wednesday, March 27
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Process systems engineering (PSE) is a subfield of chemical engineering that aims to improve systematic decision making in the design and operation of chemical processes. Optimization plays an integral role in PSE research; hence, there has always been a strong connection between PSE and operations research (OR). In the area of optimization under uncertainty, the PSE community has worked on what is referred to as flexibility analysis since the 1970s, which closely resembles what is now widely known as robust optimization. However, surprisingly, flexibility analysis has not received any attention in the OR community. In the first part of this talk, we establish the relationship between flexibility analysis and two-stage adjustable robust optimization (ARO). We further apply ARO methods to develop more efficient solution approaches to flexibility analysis problems for linear systems. In the second part, we present an application of multistage ARO using affine decision rules. Here, we consider the scheduling of power-intensive processes that can provide interruptible load, which refers to backup load reduction capacity that can be sold to provide additional revenue. However, providing interruptible load introduces uncertainty into the system as one does not know in advance when and how much load reduction will be requested.

Bio:

Qi Zhang is an Assistant Professor in the Department of Chemical Engineering and Materials Science at the University of Minnesota. He received his Ph.D. in Chemical Engineering from Carnegie Mellon University, and worked at BASF in Germany and Houston prior to joining the University of Minnesota. His research lies at the intersection of chemical engineering and operations research, primarily focused on developing optimization models and methods for the design of energy and process systems, production planning and scheduling, and supply chain optimization.