Events

"Service Systems with Server Compatibilities and Redundancy"

Presentation by Rhonda Righter
Professor
Department of Industrial Engineering and Operations Research
University of California, Berkeley

Wednesday, December 11
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

In large service systems, such as cloud computing systems, there are different classes of jobs and of servers such that each job class can only be done on a subset of the server classes, due to data locality and other constraints. Moreover, redundancy may be permitted, meaning that copies of jobs are sent upon arrival to all compatible servers. Redundant copies may be canceled once any copy begins service (cancel-on-start) or when any copy finishes service (cancel-on-completion). Under Markovian assumptions, the steady-state distributions for such systems have been shown to have a simple, “product-form” structure. I will describe a unified framework for both the cancel-on-start and cancel-on-complete models that provides a common simple proof for the product-form results at a detailed state description and provides a simple, state-aggregated, view for analyzing waiting time distributions. I will also explore conditions under which one redundancy cancellation protocol is better than the other. Joint work with Ivo Adan, Igor Kleiner, Kristen Gardner, and Gideon Weiss.

Bio:

Rhonda Righter is a Professor and past Chair of the Department of Industrial Engineering and Operations Research at the University of California, Berkeley, where she earned her PhD. Before joining the Berkeley faculty she was a Professor of Operations Management and Information Systems in the Leavey School of Business at Santa Clara University. Her primary research and teaching interests are in the general area of stochastic modeling and optimization, especially as applied to service, manufacturing, computer communication, and cloud computing systems. She has won awards for her teaching, research, and service, and is currently an associate editor for the Journal of Scheduling, Queueing Systems, Naval Research Logistics, Stochastic Models, and the INFORMS Service Science Journal. She was the founding Chair of the Applied Probability Society of INFORMS.

"Geospatial Big Data Analytics for Business and Conservation"

Presentation by Kristopher Purens
Descartes Labs
 

Wednesday, November 20
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

What is the role of Big Data for sustainability? With a background in paleontology, environmental science, and mass extinctions, Kristopher will talk about how corporate scientists can use geospatial analytics to help companies find sustainable solutions, while also helping the organizations become more efficient. With a rapid increase in the number of Earth observation satellites in the past decade, geospatial analysis is entering a new era where Big Data tools are necessary to analyze and create value from this data. From classic use cases of crop monitoring, new satellites sensor types are empowering organizations to understand carbon cycle, pollutant emissions, see through clouds, and more. 

Bio:

With a PhD in paleontology, Kristopher was on the forefront of using machine learning to fuse paleoenvironmental and life history data to understand ancient mass extinctions and climate catastrophes. Kristopher took that skill set to Shell Oil, where he worked as a data scientist to discover new petroleum deposits, before moving to Minnesota where he worked in the General Mills center of excellence for Data Science, seeking to find ways to improve their supply chain processes. Now, Kristopher is at Descartes Labs, a top geospatial startup, where he is working to combine his interests in conservation, efficiency, and natural science. 

"If You Can Send a Truck Anywhere, Where Would You Send It?"

Presentation by Michael Chmutov
C.H. Robinson
 

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

About:

As one of the world’s largest third-party logistics providers, C. H. Robinson continues to manage and optimize our customers’ supply chain solutions. As the demand for automation and streamlined decision making continues to grow from our customers, so does the adoption of new technologies and practices at C. H. Robinson. From automatic bidding to digital freight matching, our data science group has led several initiatives to change the way we move freight. In this talk, we will discuss some of our current initiatives while diving deeper into the challenges of operating a semi-dedicated fleet of trucks from a third-party logistics perspective.

Bio:

Dr. Chmutov received his Ph.D. in Mathematics, specializing in abstract algebra and algebraic combinatorics, from the University of Michigan in 2014. He completed an NSF Postdoctoral Fellowship in the Math Department at the University of Minnesota in 2018. He then decided to transition to industry, and after a brief stint as an independent contractor doing deep learning, he joined C. H. Robinson in Dec. 2018 as a Data Scientist; he now works on various routing problems.

"Freight Railway Operations Research:  A look at the most important O.R. applications in the past 20 years"

Presentation by Marc Meketon
Vice President
Oliver Wyman

Wednesday, November 13
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

The freight railways in North America have been actively using Operations Research for strategic and operational goals over the past 30 years, with some major successes over the past 20 years.  Several Franz Edelman awards (both first place, and finalist) have been given to these efforts.  This talk discusses some of the top achievements and delves lightly into the modelling.  It also describes some of the higher profile ‘failures’ where the most sophisticated approaches have not realized their potential and have been abandoned by the railways.

Bio:

Marc Meketon, a vice president in Oliver Wyman’s Transportation Practice, specializes in the design and development of software applications for optimization and analysis of transportation and logistics plans. He has developed applications ranging from optimization of railway operating plans to freight revenue management optimization to airline maintenance. Prior to Oliver Wyman, he led the operations research activities of Conrail for several years, managed a similar group at U.S. Airways, and managed airline fleet assignment research and development at AT&T Bell Laboratories.

Meketon has published papers on statistical analysis of simulation output, linear programming, communications, and simulation optimization. Two of his papers received top awards in their field. He has served on the board of the Railroad Applications Section, as well as president of the local Philadelphia chapter of INFORMS. He holds one patent. In 2003, he won the Franz Edelman Prize for Achievement in Operations Research and Management Science for helping Canadian Pacific save over $300 million. In 2016, won the Oliver Wyman Innovation Competition for his work on aircraft end-of-life planning. Meketon has a B.S. with honors in mathematics from Villanova University, and an M.S. and Ph.D. in operations research from Cornell University.

"CORTX - Optimization as a service"

Presentation by Abhishek Roy
Senior Data Scientist
Cargill
 

Wednesday, November 6
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

Cargill's Feed team provides "Feed as a Service.” This means that the team has to move 3 million tons of feed from corn milling plants to some of the world's largest feed customers across the Midwest and Texas in the most efficient way. There is just one rule: "Never run out of feed." The feed team must follow this rule while managing its systems for variable demand, variable supply, and drivers. They want to keep their customers at a target inventory while accounting for variable feed usage, shrinkage and degradation. Corn milling assets provide them with the raw material to prepare feed which can vary depending on the milling business throughput. And, they want to maximize the drivers’ time to do maximum deliveries in the shortest time window.

The objective is to decouple the feed business from the variability in different components of their supply chain. To accomplish that, Roy's data science team in collaboration with Digital Labs has broken down the problem into different optimization modules. Currently, they have finished foundational work on two modules. The KIX (Customer Inventory Execution) optimization module has two key features: Usage Model and Recommendation Model. The usage model ingests driver estimated inventory readings across all the 130 feed customers to calculate the current, and customer feed inventories and usage rates. The recommendation optimization model takes the usage model numbers to create a recommended loads of deliveries for next seven days. This list can range from 300-350 truckloads and 70 different customers per day to optimally fulfill the most urgent customer's needs considering different constraints such as customer closed days and gate hours, target inventory and plant capacity.

Bio:

Abhishek Roy has approximately 10 years of consulting experience in data analytics, data strategy, and machine learning and has worked on many data science projects across industries such as consumer packaged goods, retail, insurance, finance and manufacturing. He also co-founded the Social Data Science group, which works with nonprofits such as Generation NEXT, Avivo, and the Science Museum of Minnesota to apply data science for social good.

Roy has a master's in predictive analytics from Northwestern University in Chicago and is currently working as a senior data scientist at Cargill. In his free time, he enjoys playing basketball, cricket, and badminton.

"Integrating Anticipative Replenishment-Allocation with Reactive Fulfillment for Online Retailing Using Robust Optimization"

Presentation by Professor Yun Fong Lim
Lee Kong Chian School of Business
Singapore Management University
 

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

About:

The aggressive expansion in e-commerce sales significantly escalates online retailers’ operating cost. In each period of a planning horizon, an online retailer decides on how much to replenish each product and how to allocate its inventory to fulfillment centers (FCs) before demand is known. After the demand in the period is realized, the retailer decides on which FCs to fulfill it. It is crucial to optimize the replenishment, allocation, and fulfillment decisions jointly such that the expected total operating cost is minimized. The problem is challenging because the replenishment-allocation is done in an anticipative manner under a “push” strategy, but the fulfillment is executed in a reactive manner under a “pull” strategy. We propose a multi-period stochastic optimization model to delicately integrate the anticipative replenishment-allocation decisions with the reactive fulfillment decisions such that they are determined seamlessly as the demands are realized over time. We develop a two-phase approach based on robust optimization to solve the problem. The first phase decides whether different products should be replenished in each period (binary decisions). We fix these binary decisions in the second phase, where we determine the replenishment, allocation, and fulfillment quantities. Numerical experiments suggest that our approach outperforms existing methods from the literature in solution quality and computational time, and performs within 7% of a benchmark with perfect information. A study using data from a major fashion online retailer in Asia suggests that the two-phase approach can potentially reduce the retailer’s current cost by 30%. By decoupling the binary decisions from the continuous decisions, our methodology can handle large problem instances (up to 1,200 products). The integration, robustness, and adaptability of the decisions under our approach create significant values.

Bio:

Yun Fong LIM is Associate Professor of Operations Management at the Lee Kong Chian School of Business, Singapore Management University (SMU). He is also Chang Jiang Chair Professor, and has been a Lee Kong Chian Fellow and an NOL Fellow. Yun Fong’s research appears in Operations Research, Management Science, Manufacturing and Service Operations Management, and Production and Operations Management. He has delivered keynote and plenary speeches in several international conferences. In addition, his work has received funding by MOE and A*STAR and media coverage by The Business Times, Channel 8, and CNA938. His current research interests include e-commerce and marketplace analytics, inventory management, warehousing and fulfillment, flexible workforce and resource management, and sustainable urban logistics.

Yun Fong is a recipient of the SMU Teaching Excellence Innovative Teacher Award. He teaches both undergraduate and postgraduate courses in Operations Management. He has provided consulting service and executive development to corporations such as Maersk, McMaster-Carr Company, Resorts World Sentosa, Schneider Electrics, Temasek Holdings, and Zalora. Yun Fong obtained both his PhD and MSc degrees in Industrial and Systems Engineering from the Georgia Institute of Technology.

"The Shortest Cycle Problem and the Second Shortest Path Problem"

Presentation by Professor James Orlin
Operations Research
MIT Sloan School
 

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

About:

Professor Orlin will present an algorithm for finding the fastest algorithm for finding a shortest cycle in a directed graph. The new algorithm is actually simpler than previous fast algorithms for the shortest cycle problem. He will also address the following question: which is intrinsically an easier problem: the problem of finding the shortest cycle in a graph or the problem of finding the second shortest path from a given origin to a given destination. Professor Orlin will also present results that show that one of these two problems is a special case of the other, and is likely to be intrinsically easier than the other. 

Bio:

James Orlin is the E. Pennell Brooks Professor of Operations Research at the MIT Sloan School.   He is best known for his research on obtaining faster algorithms for problems in network and combinatorial optimization and for his text with Ravi Ahuja and Tom Magnanti entitled Network Flows: Theory, Algorithms, and Applications. He has won various awards for his co-authored publications: the 1993 Lanchester Prize for the best publication in O.R, the 2004 EXPLOR Award for leadership in online marketing research, the 2007 INFORMS Computing Society Prize for research in the interface of O.R. and computer science, the 2008 IEEE Leonard G. Abraham Prize for research in communication theory, the 2008 INFORMS Koopman Prize for research in military operations research, the 2011 IEEE Bennett Prize for research in communication theory, and the 2016 ACM SIGecom Test of Time Award, for a paper published between 10 and 25 years ago that has had “significant impact on research or applications that exemplify the interplay of economics and computation.”

"Data-driven Greedy Policies and Information-Relaxation Bounds for Ambulance Location and Deployment"

Presentation by Professor Lavanya Marla
Industrial and Enterprise Systems Engineering
University of Illinois at Urbana-Champaign
 

Wednesday, October 16
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

We present an efficient data-driven computational solution and bounding approach for static allocation of an ambulance fleet and its dynamic redeployment, where the goal is to position (or re-position) ambulances to bases to maximize the system's service level. Central to our approach is a discrete-event simulator to evaluate the impact of ambulance deployments to logs of emergency requests. We first model ambulance allocation as an approximately-submodular-maximization problem, and devise a simple and efficient greedy algorithm that produces both static allocations and dynamic repositioning policies. In parallel, we find data-driven information-relaxation bounds for both static and dynamic cases. We build even tighter information-relaxation bounds by penalizing the previous relaxations. Our approach allows the computation of tight bounds without incurring the curse of dimensionality common to such approaches. Our bounding methods help inform policymakers about the viability of proposed fleet sizes and policies being adopted by the contracted EMS agencies. Our computational experiments on an Asian city's EMS demonstrate the tractability and efficiency of our greedy algorithm and our bounding methods.
The first part of this work is with Ramayya Krishnan and Yisong Yue, and the latter part with Achal Bassamboo.

Bio:

Lavanya Marla is an Assistant Professor in Industrial and Enterprise Systems Engineering at the University of Illinois at Urbana-Champaign. Prior to her current position, she was a Systems Scientist with the Heinz College at Carnegie Mellon University; and earned her PhD in Transportation Systems from MIT and Bachelors degree from IIT Madras. Her research interests are in robust and dynamic decision-making under uncertainty and game theoretic analysis for large-scale transportation and logistics systems; combining tools from data-driven optimization, statistics, simulation and machine learning. Her research is funded by an integrative National Science Foundation grant, a Department of Homeland Security cyber-security grant, the Department of Transportation, the US-India Educational Foundation, the INFORMS Transportation and Logistics Society and aviation companies. Her work has received an Honorable mention for the Anna Valicek award from AGIFORS, a best presentation award from AGIFORS, a KDD Startup Research award, and a Top-10 cited paper recognition from Transportation Research – Part A.

"Cost-Sharing Transportation Systems"

Presentation by Professor Maged M. Dessouky
Department of Industrial and Systems Engineering
University of Southern California
 

Wednesday, October 9
3:15pm - Refreshments, Lind Hall 305
3:30pm - Graduate Seminar, Lind Hall 305

About:

A set of nascent industries focusing on cost-sharing transportation systems such as ridesharing/carsharing have recently emerged. These types of cost-sharing transportation systems are also being introduced in freight delivery through horizontal cooperation of their logistic systems to reduce costs and delay times. Horizontal cooperation achieved through pooling of freight transportation networks reduces total shipping costs, and alleviates the impact on traffic congestion. One major impediment for successful implementation of these types of transportation systems is the determination of the cost-share amount for each participant. The cost-sharing problem has largely been neglected in the literature and is the focus of this talk. One crucial component of a cost sharing transportation system is the allocation of costs and/or savings to each participant in the system. Without a model to allocate costs and/or savings to each participant in the system, there is no basis to allocate the costs in a fair manner to the participants, thus making it less of an incentive to participate. In this talk, Dessouky gives two examples of models, one for ridesharing and the other for freight consolidation, for determining the cost-share of each participant.

Bio:

Maged M. Dessouky is a Dean's Professor and Chair in the Daniel J. Epstein Department of Industrial and Systems Engineering. His research area is transportation system optimization where he has authored over 100 refereed publications. His paper “Optimal Slack Time for Schedule Based Transit Operations” was awarded the INFORMS Transportation Science and Logistics Best Paper Prize. He is a Fellow of IISE and serves as Associate Director of METRANS, a center focused on solving important urban transportation problems. He is currently area/associate editor of Transportation Research Part B: Methodological, IISE Transactions, and Computers and Industrial Engineering, on the editorial board of Transportation Research Part E: Logistics and Transportation Review, and previously served as area editor of the ACM Transactions of Modeling and Computer Simulation and associate editor of IEEE Transactions on Intelligent Transportation Systems. He received his Ph.D. in Industrial Engineering from the University of California, Berkeley, and M.S. and B.S. degrees from Purdue University.