Anyone interested in learning about recent work at Georgia Tech and opportunities for future collaboration. For additional information or questions, please contact Joscelyn Cooper (j.cooper@isye.gatech.edu).
Turgay Ayer
Research Director of the Medical Decision-Making, Center for Health and Humanitarian Systems
Jon Duke
Director, Health Analytics and Informatics, College of Computing
Sherry Farrugia
Managing Director, Health Research Partnerships, Institute for People and Technology
Pinar Keskinocak
Director, Center for Health and Humanitarian Systems
Julie Swann
Affiliate, Center for Health and Humanitarian Systems
Margaret Wagner-Dahl
Associate Vice President for Health IT, Enterprise Innovation Institute
The event is sponsored by the Georgia Tech Center for Health & Humanitarian Systems and the Stewart School of Industrial & Systems Engineering. Organizers include Georgia Tech's Enterprise Innovation Institute (EI2), Institute of People and Technology (IPAT), and the College of Computing.
]]>We strongly encourage students to act now to seek full-time employment, internships, and projects (rather than waiting until the end of the semester). Plus, enjoy a free pizza lunch!
Where: ISyE Main Bldg, 2nd Floor Atrium
When: Thursday, January 19th | 11:15am - 2:15pm
What: The session will include:
Please plan on staying for the duration of the event and bring copies of your resume and business cards. Dress is business casual.
EVENT SPONSOR
The event is sponsored through the generosity and support of APICS - Atlanta Chapter. APICS is a non-profit educational organization addressing operations management and supply chain management issues, and providing professional development opportunities to its members. APICS Membership is free for full time students. Join today at www.apics.org/join and start networking at local APIC Atlanta events. Also make sure to stop by the APICS table at the event.
]]>]]>
Relief requirements for public health and humanitarian events are in general both unknown in size and type, and are affected by dynamic and hard to measure factors such as geographic location, local economy, infrastructure, social and political conditions, etc. Preparing for long-term development and response to emergent events often involves uncertainty in timing, scope, or scale. Pre-planning for these situations requires an understanding of forecasting, distribution network design, and strategies for managing the uncertainty. This course will examine methods and models for making pre-planning decisions and explore the significant value that is obtained through informed decision-making in advance of an unpredictable event.
Tactical and strategic members of non-governmental organizations and private corporations involved in the humanitarian relief efforts, U.S. federal government employees, U.S. state or local government employees, humanitarian foundation members, program managers, directors of field operations, disaster relief managers, director of logistics and inventory, and logistics officers
NOTE: Pre-course activities will conducted online using the HELIX online learning management system. Access instructions will be provided to registrants when details become available.
Day 1
Day 2
Provided
Recommended
STUDENT REQUIREMENTS
Students need a laptop with Microsoft Excel and the ability to connect to a high-speed internet connection (internet access is provided for onsite portions of course).
For those interested in earning the Health and Humanitarian Supply Chain Management Certificate, this course is the first of the three-course certificate program. To earn the certificate, participants must register and complete the following courses within three years:
Julie Swann, Ozlem Ergun, Pinar Keskinocak
]]>
Despite having common goals, often the lack of cooperation and coordination between humanitarian organizations results in procurement and allocation inefficiencies. As a result, a systems view of a humanitarian effort is needed to ensure appropriate use of scarce resources to meet the goals at hand. This course will focus on conceptual and modeling skills to understand and effectively manage humanitarian response from a systems perspective. Models will address system characteristics (e.g. demand dependencies) that drive system dynamics and policies to regulate performance. Course topics include methods for improving coordination and collaboration, addressing demand dependencies, and reliably measuring and evaluating system performance.
Tactical and strategic members of non-governmental organizations, private corporations involved in the humanitarian relief efforts, U.S. federal government employees, U.S. state or local government employees, humanitarian foundation members, program managers, directors of field operations, disaster relief managers, director of logistics and inventory, and logistics officers
Upon completion of this course, you will be able to:
NOTE: Pre-course activities will conducted online using the HELIX platform online learning management system. Access instructions will be provided to registrants when details become available.
Coordination and Collaboration – 2 hours
Game Theory/Incentives – 1 hour
System Dynamics - 2 hours
Day 1
Day 2
Day 3
Provided
Recommended
For those interested in earning the Health and Humanitarian Supply Chain Management Certificate, this course is the third and final of the three-course certificate program. To earn the certificate, participants must register and complete the following courses within three years:
Julie Swann, Ozlem Ergun, Pinar Keskinocak
]]>Numerous tactical decisions must be made in the response to a public health or humanitarian event. Many of these decisions are concerned with the timely and efficient procurement, allocation and distribution of resources (e.g. funds, supplies, volunteers) through a supply chain. This course will explore methodologies for tactical decision making, including procurement and inventory policies, strategies for distribution and allocation of limited resources and transportation decisions.
Tactical and strategic members of non-governmental organizations, private corporations involved in the humanitarian relief efforts, U.S. federal government employees, U.S. state or local government employees, humanitarian foundation members, program managers, directors of field operations, disaster relief managers, director of logistics and inventory, and logistics officers
NOTE: Pre-course activities will conducted online using the HELIX platform online learning management system. Access instructions will be provided to registrants when details become available.
Inventory – 2 hours
Resource Allocation – 1 hour
Day 1
Day 2
Day 3
Provided
Recommended
Students need a laptop with Microsoft Excel and the ability to connect to a high-speed internet connection (internet access is provided for onsite portions of course).
For those interested in earning the Health and Humanitarian Supply Chain Management Certificate, this course is the second of the three-course certificate program. To earn the certificate, participants must register and complete the following courses within three years:
Julie Swann, Ozlem Ergun, Pinar Keskinocak
]]>Energy and Greenhouse Gas Emissions in the Supply Chain
Valerie Thomas
RSVP by September 15, 2017
Companies are measuring and reporting their supply chain energy use, greenhouse gas emissions, and other environmental metrics. The first part of the talk will overview what is being measured and reported in a selection of industries - who, what, where, when, why, and how. The second part will focus on approaches for reducing these impacts, challenges in quantifying them, and methods for improvement.
Register Online for upcoming SCL IRC seminars
]]>
*To take advantage of the included lunch, you must register by the noted deadlines. If you have any questions, please email event@scl.gatech.edu.
]]>The New Frontier of Data-Driven Price Optimization
He Wang
RSVP by November 10, 2017
In practice, firms are often faced with pricing challenges including high demand uncertainty, limited inventory, and restrictions to conduct price experimentation. In this talk, I will discuss models and algorithms that combine machine learning and price optimization. The key idea of these algorithms is to use real-time sales data to improve pricing decisions. I will report simulation and field experiment results that show significant revenue improvement using these methods.
Register Online for upcoming SCL IRC seminars
]]>
*To take advantage of the included lunch, you must register by the noted deadlines. If you have any questions, please email event@scl.gatech.edu.
]]>OVERVIEW
Additive manufacturing (AM) technologies build parts directly from digital data without any specialized or custom tooling. In the AM approach, a 3-D blueprint for an item can be downloaded from the cloud, and the item can be constructed immediately on-site, using 3-D printing equipment and feedstock materials. AM eliminates multiple time-consuming and expensive steps while significantly simplifying goods transport. AM will thus eliminate or reduce multiple supply chain tiers. The global supply base and the manufacturing landscape will be dramatically impacted through the increasing industrial adoption of AM. In this session, we will discuss the evolution of AM, some recent applications, and its impact as it relates to logistics and supply chain.
Register Online for upcoming SCL IRC seminars
]]>
*To take advantage of the included lunch, you must register by the noted deadlines. If you have any questions, please email event@scl.gatech.edu.
]]>Applications of Simulation in Supply Chain Facility Analysis and Design
Dave Goldsman, Christos Alexopoulos
RSVP by August 18, 2017
Register Online for upcoming SCL IRC seminars
We will discuss the use of simulation as a tool for analyzing and improving supply chain performance. In particular, we will show how simulation can be used to (i) evaluate the effectiveness and robustness of a particular supply chain implementation, and (ii) compare the performance of competing supply chain strategies.
]]>*To take advantage of the included lunch, you must register by the noted deadlines. If you have any questions, please email event@scl.gatech.edu.
]]>
ABSTRACT:
Multistage stochastic programming problems are an important class of optimisation problems, especially in energy planning and scheduling. These problems and their solution methods have been of particular interest to researchers in stochastic programming recently. Because of the large scenario trees that these problems induce, current solution methods require random sampling of the tree in order to build a candidate policy. Candidate policies are then evaluated using Monte Carlo simulation. Under certain sampling assumptions, theoretical convergence is obtained almost surely. In practice, the convergence of a given policy requires a statistical test and is only guaranteed at a given level of confidence.
In this talk, I will present a deterministic algorithm to solve these problems. The main feature of this algorithm is a deterministic path sampling scheme during the forward pass phase of the algorithm which is guaranteed to reduce the bound gap at all the nodes visited. Because policy simulation is no longer required, there is an improvement in performance over traditional methods for problems in which a high level of confidence is sought.
BIO: Regan Baucke is a PhD Student at the University of Auckland working under the supervisors Golbon Zakeri and Anthony Downward. His work focuses on multistage stochastic programming and risk aversion. He is a member of the EPOC research group at the University of Auckland, which focuses on mathematical modeling and optimisation with the view of analyzing and improving the New Zealand electricity market.
]]>ABSTRACT:
We present an online method that estimates the final size of the branch-and-bound tree in Mixed-Integer Programming solvers. The method combines an old sampling method due to Knuth (1975) and recent work on branching by Le Bodic and Nemhauser (2017). This method is implemented in the MIP solver SCIP and its results are displayed as an extra column. This is joint work with Gleb Belov, Samuel Esler, Dylan Fernando and George Nemhauser.
BIO: Pierre Le Bodic, Lecturer at Monash University, Melbourne, Australia
]]>ABSTRACT:
Due to the trend towards decentralization and greater complexity in supply chains, companies are increasingly exposed to supply risks. Various strategies to mitigate supply risks have been developed using modeling-based approaches, including risk diversification by dual sourcing and direct investment to improve supplier performance. Yet, despite the overwhelming evidence that managerial decisions are influenced by behavioral factors particularly under risk and uncertainty, such behavioral factors are typically not considered by previous theoretical studies. In this paper, we use controlled lab experiments to evaluate the performances of dual sourcing and single sourcing with supplier improvement strategies to mitigate risks of a buyer facing suppliers with different costs and risk profiles, and develop behavioral theories to elucidate the decision-making process under supply risks more effectively. With dual sourcing, human buyers do not diversify their orders effectively (relying on a more even allocation of orders between suppliers than theory suggests) and exhibit quantity hedging behavior. To explain this phenomenon, we propose and empirically validate a behavioral theory in which human buyers choose order quantities to minimize their disutility from order allocation errors between suppliers. Human buyers use the single sourcing with supplier improvement strategy relatively effectively, despite being subject to supplier selection errors due to bounded rationality.
]]>ABSTRACT:
Optimization problems face random constraint violations when uncertainty arises in constraint parameters. Effective ways of controlling such violations include risk constraints, e.g., chance constraints and Conditional Value-at-Risk (CVaR) constraints. This talk discusses risk constraints when the distributional information of the uncertain parameters consists of moment information (e.g., mean, covariance, support) and certain structural information, for which we mention three specific examples: alpha-unimodality, log-concavity, and dominance on the tail. We find that the ambiguous risk constraints in these settings can be recast or approximated using conic constraints that facilitate computation. Finally, we demonstrate the theoretical results via case studies on power system operation and appointment scheduling.
BIO: Ruiwei Jiang is an Assistant Professor of Industrial and Operations Engineering at the University of Michigan at Ann Arbor. His research interests include stochastic optimization and integer programming. Application areas of his work include power and water systems, healthcare, and transportation systems. Recognition of his research includes the Stochastic Programming Society student paper award, the INFORMS George E. Nicholson student paper award, and the INFORMS Junior Faculty Interest Group paper award (honorable mention).
]]>
ABSTRACT: This talk will showcase the use of some of operations research methods by providing cases in analyzing healthcare systems and studying disease progression and control.
The first part of the talk will focus on breast cancer screening policies in the high-risk population. The performance of mammography is not satisfactory for women at high-risk for breast cancer. Other technologies such as ultrasound and MRI might address some of the limitations of mammography. Currently, there is no consensus on optimal use of these technologies. Çağlayan will present the optimization model they developed to identify optimal, practical and cost-effective screening policies and their main findings.
The second part of the talk will present a novel method to optimize staffing levels in complex healthcare delivery systems such as Emergency Department (ED). Keeping waiting times at acceptable levels is not only important for patient satisfaction but also a patient safety concern in EDs. However, optimizing physician staffing levels is not an easy task for a system with unscheduled time-varying arrivals, medium-to-long service times, multiple patient classes and multiple treatment stages. In this talk, Çağlayan will introduce their novel network model to tackle the physician-staffing problem in ED.
Keywords: mixed-integer linear program, disease management, queuing theory, staffing
BIO: Çağlar Çağlayan is a Ph.D. candidate in the Operations Research program at Georgia Institute of Technology. His research interests include mathematical modeling, optimization, and data- and decision-centric healthcare analytics. He utilizes a broad range of analytical methods and collaborates with healthcare providers and researchers to conduct data-driven research with methodological contributions and clinically impactful findings.
]]>ABSTRACT:
We apply standard operational principles of inflows and outflows to estimate the number of undocumented immigrants in the United States, using the best available data, including some that has only recently become available. We generate a lower bound for the number of undocumented immigrants using conservative parameter values that underestimate inflows and overestimate outflows. Our lower bound is close to 17 million, 50% higher than the most prominent current estimate of 11.3 million, which is based on survey data and thus different sources and methods. Standard parameter values generate an estimate of 22.8 million undocumented immigrants, twice as large as the current estimate.
BIO: Mohammad Fazel-Zarandi is a Postdoctoral Associate and Lecturer at the Yale School of Management. His research interests are in the areas of Public Sector Operations, Data-Driven Decision Making, Applied Probabilistic Modeling and Statistics, and Operations Strategy. His current research focuses on the application of Operations techniques to improve decision making in the public policy domain. Before joining Yale, he received his PhD from the Rotman School of Management, University of Toronto in Operations Management and his M.Sc. in Industrial Engineering from the University of Toronto.
]]>ABSTRACT:
Distributed service systems such as data-centers and multi-core processors have enabled the exponential growth of the network infrastructure that supports the Internet. From the point of view of an administrator of such systems, the objective is to provide the best possible service to the customers (fast response times), using the least possible amount of control overhead.
These systems, and many more, can be analyzed using the following abstract model: a single stream of jobs arrive as a Poisson process of rate $\lambda$ N, with $0<\lambda<1$ fixed, and are immediately dispatched to one of several queues associated with N identical servers with unit processing rate. We assume that the dispatching decisions are made by a central dispatcher endowed with a finite memory, and with the ability to exchange messages with the servers.
In this setting, we study the fundamental resource requirements (in terms of memory bits and message exchange rate), in order to drive the expected steady-state queueing delay of a typical job to zero, as N increases. We propose a certain policy and establish that it drives the delay to zero when either (i) the message rate grows superlinearly with N, or (ii) the memory grows superlogarithmically with N. Moreover, we show that any policy that has a certain symmetry property, and for which neither condition (i) or (ii) holds, results in an expected queueing delay which is bounded away from zero.
BIO: Martin Zubeldia is a PhD candidate in the Laboratory for Information and Decision Systems at MIT, advised by professors David Gamarnik and John N. Tsitsiklis. Before joining MIT, he obtained a B.S. and a M.Sc. in Electrical Engineering from Universidad ORT Uruguay, in 2012 and 2014 respectively. His primary research interests lie broadly in the fields of applied probability and queueing theory, with a special emphasis in large-scale decision systems, and the tradeoffs between information and performance in those systems.
]]>ABSTRACT:
Data science is at a crossroads. Each year, thousands of new data scientists are entering science and technology, after a broad training in a variety of fields. Modern data science is often exploratory in nature, with datasets being collected and dissected in an interactive manner. Classical guarantees that accompany many statistical methods are often invalidated by their non-standard interactive use, resulting in an underestimated risk of falsely discovering correlations or patterns. It is a pressing challenge to upgrade existing tools, or create new ones, that are robust to involving a human-in-the-loop.
In this talk, I will describe two new advances that enable some amount of interactivity while testing multiple hypotheses, and control the resulting selection bias. I will first introduce a new framework, STAR, that uses partial masking to divide the available information into two parts, one for selecting a set of potential discoveries, and the other for inference on the selected set. I will then show that it is possible to flip the traditional roles of the algorithm and the scientist, allowing the scientist to make post-hoc decisions after seeing the realization of an algorithm on the data. The theoretical basis for both advances is founded in the theory of martingales : in the first, the user defines the martingale and associated filtration interactively, and in the second, we move from optional stopping to optional spotting by proving uniform concentration bounds on relevant martingales.
BIO: Aaditya Ramdas is a postdoctoral researcher in Statistics and EECS at UC Berkeley, advised by Michael Jordan and Martin Wainwright. He finished his PhD in Statistics and Machine Learning at CMU, advised by Larry Wasserman and Aarti Singh, winning the Best Thesis Award in Statistics. A lot of his research focuses on modern aspects of reproducibility in science and technology — involving statistical testing and false discovery rate control in static and dynamic settings.
]]>
ABSTRACT:
As market design theory increasingly shapes the design and operations of real-life marketplaces, it is important for designers to provide simple policy levers that practitioners can use to optimize platform objectives. In the school choice market, where scarce public school seats are assigned to students, a key operational issue is how to reassign seats that are vacated after an initial round of centralized assignment. Practical solutions to the reassignment problem must be simple to implement, truthful, efficient and fair while also alleviating costly student movement between schools.
In this talk, I will propose and axiomatically justify a class of reassignment mechanisms, the Permuted Lottery Deferred Acceptance (PLDA) mechanisms. Our mechanisms generalize the commonly used Deferred Acceptance (DA) school choice mechanism to a two-round setting and retain its desirable incentive, fairness and efficiency properties. School choice systems typically run Deferred Acceptance with a lottery number assigned to each student to break ties in school priorities. I will show that under natural conditions on demand, correlating the tie-breaking lotteries across rounds preserves allocative welfare, and reversing the first-round lottery order minimizes reassignment among all PLDA mechanisms. Empirical investigations based on data from NYC high school admissions support our theoretical findings.
This is based on joint work with Itai Feigenbaum, Yash Kanoria and Jay Sethuraman.
BIO: Irene Lo is a 5th year Ph.D. student in the Industrial Engineering & Operations Research department at Columbia University, advised by Jay Sethuraman, Jacob Leshno and Yash Kanoria. Her main research interests are in operational and algorithmic issues in marketplace design, particularly in using theory to provide practical solutions for markets with social value. Her current focus is on operational challenges in school choice mechanisms. She is also interested in platform-based marketplaces, mechanism design for social good, graph theory, and games on networks. She graduated from Princeton University in 2013 with an A.B. in mathematics.
]]>ABSTRACT:
In the first part of the talk, we provide an overview of recent and ongoing work in the public health domain. Specifically, we propose a social network based approach for preventing drug abuse among homeless youth, a robust optimization technique for preventing suicide among freshmen, and a mixed integer optimization approach for designing fair, efficient, and interpretable policies for allocating scarce resources (e.g., kidneys for transplantation, houses for homeless youth).
In the second part of the talk, we focus on a specific project. We present a data-driven optimization approach to estimate wait times for individual patients in the U.S. Kidney Allocation System, based on the very limited system information that they possess in practice. To deal with this information incompleteness, we develop a novel robust optimization analytical framework for wait time estimation in multiclass, multiserver queuing systems. We calibrate our model with highly detailed historical data and illustrate how it can be used to inform medical decision making and improve patient welfare.
The first part of the talk is joint work with Milind Tambe, Eric Rice, and our students and fellows of the Center for AI in Society. The second part of the talk is joint work with Chaitanya Bandi and Nikolaos Trichakis.
BIO: Phebe Vayanos is an Assistant Professor of Industrial & Systems Engineering and Computer Science at the University of Southern California. She is also an Associate Director of the CAIS Center for Artificial Intelligence in Society at USC. Her research interests include optimization under uncertainty, data-driven optimization and analytics, artificial intelligence, and machine learning. Her research is motivated by real problems that are important for society. Prior to joining USC, she was lecturer in the Operations Research and Statistics Group at the MIT Sloan School of Management, and a postdoctoral research associate in the Operations Research Center at MIT. She holds a PhD degree in Operations Research and an MEng degree in Electrical & Electronic Engineering, both from Imperial College London.
]]>
ABSTRACT:
The recent (r)evolution of “transportation-as-a-service” has affected commuting patterns in major American cities. Yet the rise of ride-sharing apps, like Uber or Lyft, and bike-sharing systems, like CitiBike or Hubway, not only provides new opportunities for commuters, but also new challenges for operators. Common to all of these challenges are the intricate underlying network effects each ride has on supply in the system. For example, every rental of a bike at a bike-sharing station not only decreases the supply of bikes at that station but simultaneously increases the supply of docks available (for bike returns). A similar phenomenon is present in ride-sharing. The resulting externalities, positive or negative, are of both academic and practical interest.
In this talk, I present the results of two orthogonal pieces of work that combine rigorous mathematical analysis with real data to address these operational challenges. In the first part, we apply an inventory model frequently used in routing problems to inform the system-design of bike-sharing systems. By identifying an underlying discrete convexity in the model, we develop a provably correct, fast optimization algorithm. Applying our algorithm to data-sets from NYC, Chicago, and Boston, we derive proposals for re-designing these systems, which have since been adopted by the operators. In the second part, we study the question of how to optimize prices for ride-sharing systems in the presence of network externalities. Though the underlying stochastic control problem is non-convex, we show that a novel relaxation can be efficiently solved and provides parametric approximation guarantees with relative error bounds close to 0 in realistic regimes. Surprisingly, our analysis extends far beyond the realm of pricing, unifying several results on other controls employed in such systems, which were obtained concurrently.
This talk presents two papers, the first joint with Shane G. Henderson & David B. Shmoys and the second joint with Siddhartha Banerjee & Thodoris Lykouris.
BIO: Daniel Freund is a Ph.D. Candidate in Cornell’s Center for Applied Mathematics where he is advised by Prof. David B. Shmoys. He holds a B.Sc. in Mathematics from the University of Warwick and is an alumnus of the German National Merit Foundation. His research considers optimization problems that arise at the intersection of on-demand transportation and the sharing economy. During his Ph.D., he spent time as a Data Scientist both at Motivate, the operator of America’s largest bike-sharing systems, and at Lyft. While embedded in industry, he developed tools to cope with the operational challenges arising in such transportation systems; these motivated the theoretical models and novel algorithmic advances in his thesis, which in turn had impact on real-world decision-making. He also aims to bring his industry experience to the classroom, having taught as an instructor in Cornell’s School of Operations Research and Information Engineering and received a Yahoo! Graduate Teaching Award for his role as a Teaching Assistant in the Department of Computer Science.
]]>Guests from Microsoft, McKesson, Ernst & Young, HD Supply, InterContinental Hotels Group, and Habasit will introduce their roles and engage participants in a mini-case to showcase how project management plays a huge role in several different areas.
RSVP at this link by 11:59pm by Monday, November 6, 2017 to confirm your attendance: https://goo.gl/forms/CPU9LQMtBQNcVCgA2
Attend our event if you would like to:
Please email your resume to apics.gt@gmail.com by 11:59pm on Monday, November 6, 2017 to be included in our resume book which will be distributed to respective companies.
We hope to see you there!
Please don't hesitate to email apics.gt@gmail.com or send us a Facebook message if you have any questions.
Join us to network and hear from major manufacturers about their experience with IoT. We will be joined by representatives from AGCO, Air Force Research Laboratory, BMW, EATON, EXIDE Technologies, Georgia Tech Research Institute, Intel, National Instruments, Rockwell Automation, Shaw Floors and Universal Robots.
Visit http://ws17.iotfm.org for more information and to register online.
Hosted by the Georgia Tech Manufacturing Institute. We bring together top researchers and thought leaders from the many varied disciplines that shape manufacturing — science, engineering, policy, robotics, and management — to help define and solve some of the greatest challenges facing U.S. industry today:
ABSTRACT:
We develop a novel, data-driven approach to deal with appointment sequencing and scheduling in a multi-server system, where both customer punctuality and service times are stochastic. Our model is calibrated using a data set of unprecedented resolution, gathered at a large-scale outpatient oncology practice. This data set combines real-time locations, electronic health records and appointments logs. Our approach yields tractable and scalable solutions that accommodate hundreds of jobs and servers. We demonstrate the performance of our algorithm by comparing it with existing state-of-the-art sequencing and scheduling algorithms.
BIO: Petar Momcilovic is an associate professor in the Department of Industrial and Systems Engineering at the University of Florida. His research interests are in the domain of stochastic modeling and applied probability. He received the PhD degree in Electrical Engineering from Columbia University. His research has been supported by NSF, NIH and IBM Research.
]]>Scott Luton from APICS Atlanta will join the event. Scott serves as Managing Partner for TalentStream, was a past APICS Atlanta president and is currently a member of the APICS Southeast District Staff.
For more details, please visit our event Facebook page.
]]>We strongly encourage students to act now to seek full-time employment, internships, and projects (rather than waiting until the end of the semester). Plus, enjoy a free pizza lunch!
Where: ISyE Main Bldg, 2nd Floor Atrium
When: Wednesday, November 8 | 11:00am - 2:30pm
What: The session will include:
Please plan on staying for the duration of the event and bring copies of your resume and business cards. Dress is business casual.
EVENT SPONSOR
The event is sponsored through the generosity and support of JP Morgan Chase & Co. and APICS - Atlanta Chapter. APICS is a non-profit educational organization addressing operations management and supply chain management issues, and providing professional development opportunities to its members. APICS Membership is free for full time students. Join today at www.apics.org/join and start networking at local APIC Atlanta events. Also make sure to stop by the APICS table at the event.
]]>]]>
ABSTRACT:
One of the paramount challenges of this century is that of understanding complex, dynamic, large-scale networks. Such high-dimensional networks, including social, financial, and biological networks, cover the planet and dominate modern life. In this talk, we propose novel approaches to inference in such networks, for both active (interventional) and passive (observational) learning scenarios. We highlight how timing could be utilized as a degree of freedom that provides rich information about the dynamics. This information allows resolving direction of causation even when only a subset of the nodes is observed (latent setting). In the presence of large data, we propose algorithms that identify optimal or near-optimal approximations to the topology of the network.
BIO: Negar Kiyavash is Willett Faculty Scholar at the University of Illinois and a joint Associate Professor of Industrial and Enterprise Engineering (IE) and Electrical and Computer Engineering (ECE). She is the director of Advance Data Analytics Program in IE and is further affiliated with the Coordinated Science Laboratory (CSL) and the Information Trust Institute. She received her Ph.D. degree in ECE from the University of Illinois at Urbana-Champaign in 2006. Her research interests are in design and analysis of algorithms for network inference and security. She is a recipient of NSF CAREER and AFOSR YIP awards and the Illinois College of Engineering Dean's Award for Excellence in Research.
]]>
ABSTRACT:
High-dimensional data arise in many fields of contemporary science and introduce new challenges in statistical learning and data recovery. Many data sets in image analysis and signal processing are in a high-dimensional space but exhibit a low-dimensional structure. We are interested in building efficient representations of these data for the purpose of compression and inference, and giving performance guarantees depending on the intrinsic dimension of data. I will present two sets of problems: one is related with manifold learning; the other arises from imaging and signal processing where we want to recover a high-dimensional, sparse vector from few linear measurements. In the first problem, we model a data set in $R^D$ as samples from a probability measure concentrated on or near an unknown $d$-dimensional manifold with $d$ much smaller than $D$. We develop a multiscale adaptive scheme to build low-dimensional geometric approximations of the manifold, as well as approximating functions on the manifold. The second problem arises from source localization in signal processing where a uniform array of sensors is set to collect propagating waves from a small number of sources. I will present some theory and algorithms for the recovery of the point sources with high precision.
BIO: Dr. Wenjing Liao is an assistant professor in the School of Mathematics at Georgia Tech. She obtained her Ph.D in mathematics at University of California, Davis in 2013, and B.S. at Fudan University in 2008. She was a visiting assistant professor at Duke University from 2013 to 2016, as well as a postdoctoral fellow at Statistical and Applied Mathematical Sciences Institute from 2013 to 2015. She worked at Johns Hopkins University as an assistant research scientist from 2016 to 2017. She works on theory and algorithms in the intersection of applied math, machine learning and signal processing. Her current research interests include multiscale methods for dimension reduction, regression on data, convex and non-convex optimization, source localization and sensor calibration in signal processing.
]]>ABSTRACT:
Many manufacturing firms use aggregated data to provide scheduling/decision solutions for handling their daily operations. Given the nature of shop floor operating in real-time, these average-based scheduling systems cannot be fully executed since unexpected events will almost always occur such as rush orders, design changes, machine breakdowns, defective parts, and delivery delays etc. Currently, shop-floor responds to unexpected events via manually scheduling or by Excel, which leads to poor predictability and visibility of performance, slow response to uncertainties and market changes, and low efficiency of their production and supply chain systems.
In this talk, Manufacturing Execution Optimization (MEO) technologies developed by Dr. Shi and her team will be presented. MEO aims to bridge the gap between the top-level management data typically from ERP systems and the shop-floor operations. By establishing top floor to shop floor communication, manufacturing firms will be able to significantly improve their production and supply chain efficiency while achieving a faster response to changes and disturbances in the most time-optimal manner. MEO is developed based on Nested Partitions (NP) optimization framework. The coordination nature of the NP framework provides an efficient and effective platform for information sharing and exchange in real time. In this talk, two new simulation optimization methods will also be discussed and a case study will be presented.
BIO: Leyuan Shi Professor in the Department of Industrial and Systems Engineering at University of Wisconsin-Madison and also the founding chair of the Department of Industrial Engineering and Management at Peking University of China. She received her Ph.D. in Applied Mathematics from Harvard University in 1992. Her research interests include simulation modeling and large-scale optimization with applications to operational planning and scheduling and digital supply chain management. She has developed a novel optimization framework, the Nested Partitions Method that has been applied to many large-scale and complex systems optimization problems. Her research work has been funded by NSF, NSFC, NIH, AFSOR, ONR, MOST of China, State of Wisconsin, and many private industrial companies with a total funding of more than 15 million dollars. Shi has published 3 books and more than 130 papers. She is currently serving as Editor for IEEE Trans on Automation Science and Engineering and had served on the editorial board for Manufacturing & Service Operations Management and INFORMS Journal on Computing. She was General Chair, co-Chair, and program committee for many national and international conferences. She is also one of the inventors for a set of digital tools including Manufacturing Execution Optimization (MEO), Maintenance Repair & Overhaul Optimization (MRO2), and Dynamic Manufacturing Critical-Path Time (DMCT). She is an IEEE Fellow.
]]>He will be talking about various programs at NSF CISE including core programs and cross cutting programs. For the benefit of junior faculty, I will be covering trends and tips in CAREER as well as CRII programs. For the core programs, the focus would be on Algorithmic Foundations and applied areas within the AF fold. We shall see how this program interfaces with Robust Intelligence for Machine Learning Theory, with Engineering for topics related to Optimization, with Mathematics for topics related to Numerical Computing, with Biology and with Quantum Architecture.
BIO: Dr. Rahul Shah is a Roy Paul Daniel Distinguished Associate professor at Louisiana State University (LSU). He is currently serving as a program director in Algorithmic Foundation (AF) program in National Science Foundation. He did is undergraduate degree from Indian Institute of Technology, Bombay and his Doctoral degree in computer science from Rutgers University. His main areas of interest are data structures and algorithms – particularly for string matching. Prior to joining LSU, he was an assistant research professor at Purdue University.
]]>
The TRIAD Kickoff event includes a panel discussion featuring TRIAD leadership and remarks from special guests including Executive Vice President for Research Stephen Cross and Provost and Executive Vice President for Academic Affairs Rafael Bras, as well as talks from participating faculty members.
Buffet lunch served at 11 am.
Please RSVP: http://bit.ly/2htKyLx
]]>
]]>
ABSTRACT:
We are interested in using Markov chain methods to establish convergence in probability for various algorithms in dynamic programming and optimization. We start by investigating simple "empirical" variants of classical value and policy iteration for dynamic programming. In this case, we show that the progress of these algorithms is stochastically dominated by an easy to analyze Markov chain, from which we can extract a convergence rate for the original algorithms. We continue by showing that this same line of reasoning covers several empirical algorithms in optimization as well. We argue that the advantage of this approach lies in its simplicity and intuitive appeal.
]]>
ABSTRACT:
In this talk I will give a brief overview of current academic research on Fintech by using tools from operations research and statistics. The topics to be discussed include: (1) P2P equity financing: how to design contracts suitable for a P2P equity financing platform with information asymmetry. (2) Robotic financial advising: how to get investor’s risk aversion parameters automatically by asking simple questions, and how to get consistent answers to meet goals of investors, such as retirement planning. (3) Economics of Bitcoin: how to build a general equilibrium model for bitcoin. (4) Data privacy preservation: how to do econometrics based on the encrypted data while still preserving privacy. All the above 4 topics are based on my recent working papers..
BIO: Steven Kou is a Class ‘62 Chair Professor of Mathematics and the Director of the Risk Management Institute at the National University of Singapore. Previously, he taught at Columbia University (from 1998 to 2014), University of Michigan (1996-1998), and Rutgers University (1995-1996). He teaches courses in quantitative finance, stochastic models, and statistics. Currently he is a co-area-editor for Operations Research, and has served on editorial boards of many journals, such as Management Science, Mathematical Finance, Advances in Applied Probability, Mathematics of Operations Research. He won the Erlang Prize from INFORMS in 2002. Some of his research results have been incorporated into standard MBA textbooks and have implemented in commercial software packages and terminals, e.g. in Bloomberg Terminals.
]]>
ABSTRACT: Transportation is entering an era of unprecedented transformation, with connectivity and automation in the vehicle revealing new possibilities for innovation to enable societal impact. In this context, vehicular traffic congestion in a vehicle-to-vehicle (V2V) communication environment can lead to congestion effects for information flow propagation. Such congestion effects can impact whether a specific information packet of interest can reach a desired location, and if so, in a timely manner to influence the traffic system performance. Motivated by the usefulness and timeliness of information propagation, this talk aims to characterize the information flow propagation wave (IFPW) for an information packet in a congested V2V communication environment under an information relay control strategy. This strategy seeks to exclude information that is dated in the communication buffer under a first-in, first-out queue discipline, from being relayed if the information flow regime is congested. A macroscopic two-layer model is proposed to characterize the IFPW. The upper layer, inspired by some conceptual similarities with models for disease spreading in epidemiology, is formulated as integro-differential equations to characterize the information dissemination in space and time under this control strategy. The lower layer adopts the Lighthill-Whitham-Richards model to capture the traffic flow dynamics. Closed-form solutions for the asymptotic IFPW speed and asymptotical density of informed vehicles are derived under homogeneous traffic conditions, and numerical solutions are illustrated for heterogeneous conditions. The study insights can be leveraged to develop a new generation of information dissemination strategies, and to determine V2X-infrastructure locations.
BIO: Dr. Srinivas Peeta is the Jack and Kay Hockema Professor in Civil Engineering at Purdue University. He is the Director of the NEXTRANS Center, the U.S. Department of Transportation’s (USDOT’s) Region 5 Regional University Transportation Center (UTC) (2006-2018). He is also the Associate Director of the USDOT Center for Connected and Automated Transportation, the new USDOT Region 5 UTC (2016-2022). He received his B.Tech., M.S. and Ph.D. degrees from the Indian Institute of Technology (Madras), Caltech and The University of Texas at Austin, respectively. He was a past Chair (2007-2013) of the Committee on Transportation Network Modeling of the Transportation Research Board of the National Academies. He has authored or co-authored over 270 articles in refereed journals and conference proceedings. His research interests are multidisciplinary and broadly span transportation and infrastructure systems.
]]>ABSTRACT: We propose bootstrap methodologies for simultaneous inference of low-dimensional parameters with high dimensional data. We focus on simultaneous confidence intervals for individual coefficients in linear regression, although our approach is applicable in much broader contexts. The problem can be solved by de-biasing regularized estimators such as the Lasso. However, the Bonferroni adjustment is overly conservative and asymptotic theory is complicated, especially for non-Gaussian and heteroscedastic errors. We propose residual, wild and empirical bootstrap methodologies for more accurate and robust simultaneous inference and study sample size requirements and other properties of such procedures. Our theory is complemented by many empirical results.
]]>
ABSTRACT:
Modern AI and machine learning techniques increasingly depend on highly complex, hierarchical (deep) probabilistic models to reason with complex relations and learn to predict and act under uncertain environment. This, however, casts a significant demand for developing efficient computational methods for handling highly complex probabilistic models for which exact calculation is prohibitive. In this talk, we discuss a new framework for approximate learning and inference that combines ideas from Stein's method, an advantaged theoretical technique developed by mathematical statistician Charles Stein, with practical machine learning and statistical computation techniques such as variational inference, Monte Carlo, optimal transport and reproducing kernel Hilbert space (RKHS). Our framework provides a new foundation for probabilistic learning and reasoning and allows us to develop a host of new algorithms for a variety of challenging statistical tasks, that are significantly different from, and have critical advantages over, traditional methods. Examples of applications include computationally tractable goodness-of-fit tests for evaluating highly complex models, new efficient approximation inference methods for scalable Bayesian computation, amortized maximum likelihood training for deep generative models, and new policy gradient methods that yield better exploration using Bayesian uncertainty for deep reinforcement learning.
BIO: Qiang Liu is an assistant professor of computer science at Dartmouth College. His research interests are in machine learning, Bayesian inference, probabilistic graphical models and deep learning. He received his Ph.D from University of California at Irvine, followed with a postdoc at MIT CSAIL. He is an action editor of journal of machine learning research.
]]>SCL managing director Tim Brown will be part of "The Future Supply Chain IS Digital" session 9/25 from 10:30am-11:45am. Hear how the notable supply chains of Atlanta are transforming from industrial to digital and discover the strategies you need to deploy to keep up with them.
To learn more about the annual conference, please visit https://cscmpedge.org/ehome/cscmp2017.
]]>ABSTRACT:
Cluster structures are ubiquitous for large data, and community detection has recently attracted much research attention in applied physics, sociology, computer science and statistics due to its broad applicability in various network datasets. A variety of approaches distinct in essence have thus been proposed, among which convex relaxation have not been extensively explored due to the lack of knowledge of its statistical advantages over other methods, either theoretical or empirical. In this talk, I will focus on explaining the benefits of convex community detection in two aspects: robustness against adversarial nodes and efficacy in networks with heterogeneous degrees. For the robustness, I will show that convex relaxation is able to detect the hidden communities in presence of a portion of arbitrary or even adversarial nodes with strong theoretical guarantees, while standard spectral clustering may fail due to a small fraction of outliers; For networks with heterogenous degrees, I will show that a convex optimization enjoys desirable theoretical properties under the degree-corrected stochastic block model as well as competitive empirical performances compared to the state-of-the-art tractable methods in the literature. The talk consists of my collaborative works with T. Tony Cai, Yudong Chen, and Jiaming Xu.
BIO: Dr. Xiaodong Li is an assistant professor in the statistics department at UC Davis. Prior to that, he worked in the statistics department of Wharton School at University of Pennsylvania for two years. He got Ph.D of mathematics at Stanford University in 2013, and BS at Peking University in 2008. He works on theory and methods in in the intersection of statistics, applied math and machine learning, with current research interests including dimension reduction, randomized algorithms, network data, convex and nonconvex optimization, etc. He has published a series of papers regarding low-rank recovery, matrix completion, phase retrieval and community detection in various journals of statistics, mathematics and engineering such as JACM, IEEE TIT, AoS, ACHA, FOCM, etc.
]]>ABSTRACT:
We propose to test the topological structures of complex networks under the graphon model. Graphon is a nonparametric model for large scale stochastic graphs. Many works have been done on graphon estimation, however it is not easy to interpret the network structures from estimators. We provide an inferential toolkit to study the persistent homology of the graphon landscape which reveals the clustering structure of stochastic networks. Our methods are applied to the neuroscience data related to visual memories.
Bio: Junwei Lu is a final year Ph.D. student in the Department of Operations Research and Financial Engineering at Princeton University. His researches focus on new inferential methods for modern statistical analysis with complex data structures and complicated algorithms. Junwei Lu has received several academic rewards, including Award for Excellence in Princeton Engineering School, the ICSA best student paper award and ASA Best Student Paper in Nonparametric Statistics.
]]>ABSTRACT:
Due to factors such as climate change, forest fire, plague of insects on lumber quality, it is important to update (statistical) procedures in American Society for Testing and Materials (ASTM) Standard D1990 (adopted in 1991) from time to time. The statistical component of the problem is to detect the change in the lower percentiles of the solid lumber strength. Verrill et al. (2015) studied eight statistical tests proposed by wood scientists to determine if they perform acceptably when applied to test data from a monitoring program. Some well-known methods such as Wilcoxon and Kolmogorov-Smirnov tests are found to have severely inflated type I errors when the data are clustered. A new method that performs well in the presence of random effects is therefore in urgent need. In this talk, we develop a novel test by combining composite empirical likelihood, cluster-based bootstrapping and density ratio model. The test satisfactorily controls the type I error in monitoring the trend of lower quantiles and conclusions are supported by asymptotic results. Our results are generic, not confined to wood industry applications.
BIO: Prof. Jiahua Chen obtained his Ph.D degree in 1990 under the supervision of Professor C.F. Jeff Wu. He joined the Department of Statistics and Actuarial Science at the University of Waterloo in Ontario, Canada until 2006. In Jan 2007, Professor Chen was appointed as Canada Research Chair, Tier I at the University of British Columbia at Vancouver. Dr. Chen has broad research interest in statistics. He earned his Ph.D degree with a thesis on the design of experiment, started work on sampling survey problem soon after, got interested in empirical likelihood under the influence of his friend, and was guided into the area of mixture models by Professor Kalbfleisch. He is also working statistical genetic problems and gets involved in a clinical trial.
]]>There is a recent surge in research activities around the idea of the so-called “deep neural networks” (DNN). As a technical item, DNN without a doubt is an important classroom topic and several tutorial articles and related learning resources are available. Nonetheless, streams of questions about DNN never subside from students or researchers and there appears to be a frustrating tendency among the learners to treat DNN simply as a black box. In this talk, a pedagogy is attempted with the aim to present DNN in the well-established traditional pattern recognition framework so that a deeper understanding of DNN can be reached through proper contrast to conventional techniques. Furthermore, the structure of DNN is in essence no different from a conventional feedforward neural network or multilayer perceptron, a fact that often triggers the perplexing question why it took sixty years to get to what it is today. In particular, we review the developmental aspect of DNN, in terms of how advances in connectionist models have evolved into this powerful technique. Time permitting, we’ll discuss the application of DNN in the area of automatic speech recognition so as to ascertain its efficacy, as compared to traditional statistical modeling, and to bring to surface possibly unrealized potentials of DNN as well as conventional techniques.
BIO: Biing-Hwang (Fred) Juang is the Motorola Foundation Chair Professor and a Georgia Research Alliance Eminent Scholar at Georgia Institute of Technology. He is also enlisted as Honorary Chair Professor at several renowned universities. He received a Ph.D. degree from University of California, Santa Barbara. He had conducted research work at Speech Communications Research Laboratory (SCRL) and Signal Technology, Inc. (STI) in the late 1970s on a number of Government-sponsored research projects and at Bell Labs during the 80s and 90s until he joined Georgia Tech in 2002. Prof. Juang’s notable accomplishments include development of vector quantization for voice applications, voice coders at extremely low bit rates (800 bps and ~300 bps), robust vocoders for satellite communications, fundamental algorithms in signal modeling for automatic speech recognition, mixture hidden Markov models, discriminative methods in pattern recognition and machine learning, stereo- and multi-phonic teleconferencing, and a number of voice-enabled interactive communication services. He was Director of Acoustics and Speech Research at Bell Labs (1996-2001).
Prof. Juang has published extensively, including the book “Fundamentals of Speech Recognition”, co-authored with L.R. Rabiner, and holds nearly two dozen patents. He received the Technical Achievement Award from the IEEE Signal Processing Society in 1998 for contributions to the field of speech processing and communications and the Third Millennium Medal from the IEEE in 2000. He also received two Best Senior Paper Awards, in 1993 and 1994 respectively, and a Best Paper Awards in 1994, from the IEEE Signal Processing Society. He served as the Editor-in-Chief of the IEEE Transactions on Speech and Audio Processing from 1996 to 2002. He was elected an IEEE Fellow (1991), a Bell Labs Fellow (1999), a member of the US National Academy of Engineering (2004), and an Academician of the Academia Sinica (2006). He was named recipient of the IEEE Field Award in Audio, Speech and Acoustics, the J.L. Flanagan Medal, and a Charter Fellow of the National Academy of Inventors (NAI), in 2014
]]>Stochastic gradient-based methods are the state-of-the-art in large-scale machine learning optimization due to their extremely efficient per-iteration computational cost. Second-order methods, that use the second derivative of the optimization objective, are known to enable faster convergence. However, the latter have been much less explored due to the high cost of computing the second-order information. We will present second-order stochastic methods for (convex and non-convex) optimization problems arising in machine learning that match the per-iteration cost of gradient descent, yet enjoy the faster convergence properties of second-order optimization.
Joint work with Naman Agarwal and Brian Bullins (ICML '16), and Agarwal, Bullins, Allen-Zhu and Ma (STOC '17)
BIO:
Elad Hazan is a professor of computer science at Princeton university. He joined in 2015 from the Technion, where he had been an associate professor of operations research. His research focuses on the design and analysis of algorithms for fundamental problems in machine learning and optimization. Amongst his contributions are the co-development of the AdaGrad algorithm for training learning machines, and the first sublinear-time algorithms for convex optimization. He is the recipient of (twice) the IBM Goldberg best paper award in 2012 for contributions to sublinear time algorithms for machine learning, and in 2008 for decision making under uncertainty, a European Research Council grant , a Marie Curie fellowship and Google Research Award (twice), and winner of the Bell Labs Prize. He serves on the steering committee of the Association for Computational Learning and has been program chair for COLT 2015.
]]>ABSTRACT:
Matrix completion is a fundamental machine learning problem with wide applications in collaborative filtering and recommender systems. Typically, matrix completion are solved by non-convex optimization procedures, which are empirically extremely successful. We prove that the symmetric matrix completion problem has no spurious local minima, meaning all local minima are also global. Thus the matrix completion objective has only saddlepoints an global minima. Next, we show that saddlepoints are easy to avoid for even Gradient Descent -- arguably the simplest optimization procedure. We prove that with probability 1, randomly initialized Gradient Descent converges to a local minimizer. The same result holds for a large class of optimization algorithms including proximal point, mirror descent, and coordinate descent.
BIO: Jason Lee is an assistant professor in Data Sciences and Operations at the University of Southern California. Prior to that, he was a postdoctoral researcher at UC Berkeley working with Michael Jordan. Jason received his PhD at Stanford University advised by Trevor Hastie and Jonathan Taylor. His research interests are in statistics, machine learning, and optimization. Lately, he has worked on high dimensional statistical inference, analysis of non-convex optimization algorithms, and theory for deep learning.
]]>We consider the problem of determining the optimal accuracy of public statistics when increased accuracy requires a loss of privacy. To formalize this allocation problem, we use tools from statistics and computer science to model the publication technology used by a public statistical agency. We derive the demand for accurate statistics from first principles to generate interdependent preferences that account for the public-good nature of both data accuracy and privacy loss. We first show data accuracy is inefficiently under-supplied by a private provider. Solving the appropriate social planner’s problem produces an implementable publication strategy. We implement the socially optimal publication plan for statistics on income and health status using data from the American Community Survey, National Health Interview Survey, Federal Statistical System Public Opinion Survey and Cornell National Social Survey. Our analysis indicates that welfare losses from providing too much privacy protection and, therefore, too little accuracy can be substantial.
BIO: John M. Abowd is Associate Director for Research and Methodology and Chief Scientist at the Census Bureau and the Edmund Ezra Day Professor of Economics, Professor of Statistics and Information Science at Cornell University. He is also Research Associate at the National Bureau of Economic Research (on leave while serving in the federal government), Research Affiliate at the Centre de Recherche en Economie et Statistique (CREST, Paris, France), Research Fellow at the Institute for Labor Economics (IZA, Bonn, Germany), and Research Fellow at IAB (Institut für Arbeitsmarkt-und Berufsforschung, Nürnberg, Germany). He is the past President (2014-2015) and Fellow of the Society of Labor Economists; past Chair (2013) of the Business and Economic Statistics Section and Fellow of the American Statistical Association; elected member of the International Statistical Institute; and a fellow of the Econometric Society. He has served as Distinguished Senior Research Fellow at the United States Census Bureau (1998-2016) and on the National Academies’ Committee on National Statistics (2010-2016). He currently serves on the American Economic Association’s Committee on Economic Statistics (2013-2018). He was the Director of the Cornell Institute for Social and Economic Research (CISER) from 1999 to 2007. His current research focuses on the creation, dissemination, privacy protection, and use of linked, longitudinal data on employees and employers.
]]>ABSTRACT:
Choice-based demand models based on the random utility maximization (RUM) principle are routinely used in academic literature and industry practice. However, the RUM principle may be violated in practice because customer preferences may not be rational. This raises the following empirical questions: (a) Given a dataset consisting of offer sets and individual choices, are the observed choice probabilities consistent with the RUM principle? (b) If not, what is the degree of inconsistency?
We formulate the problem of quantifying the limit of rationality (LoR) in choice modeling applications. Computing LoR is intractable in the worst case, but we identify the source of complexity through new concepts of rational separation and choice graph. By exploiting the graph structure, we provide practical methods to compute LoR efficiently for a large class of applications. Applying our methods to real-world grocery sales data, we identify product categories for which going beyond rational choice models is necessary to obtain acceptable performance.
Joint work with Srikanth Jagabathula (NYU)
------------------------------
BIO: Paat Rusmevichientong is a Professor of Data Sciences and Operations in the Marshall School of Business at the University of Southern California. His research interests focus on stochastic optimization problems with applications to market systems, supply chain, and revenue management. For more information, see www-bcf.usc.edu/~rusmevic
]]>ABSTRACT:
The ongoing shortage of organs for transplantation has inspired a vibrant literature on organ allocation. By contrast, organ donation has been little explored. In this paper, we develop a parsimonious model of organ donation to analyze the social-welfare consequences of introducing the donor-priority rule, which grants registered organ donors priority in receiving organs, should they need transplants in the future. We model an individual’s decision to join the donor registry, which entails a trade-off between abundance of supply, exclusivity of priority, and cost of donating (e.g., psychological burden). By incorporating heterogeneity in the probability of requiring an organ transplant and in organ quality, we show that, in contrast to the literature, introducing the donor-priority rule can lower social welfare due to unbalanced incentives across different types of individuals. In view of the potentially undesirable social-welfare consequences, we propose a freeze-period remedy, under which an individual is not entitled to a higher queueing priority until after having been on the organ-donor registry for a specified period. We show that, echoing the theory of the second best (Lipsey and Lancaster 1956), this additional market friction helps rebalance the incentive structure, and in conjunction with the donor-priority rule, can guarantee an increase in social welfare by boosting organ supply without compromising organ quality or inducing excessively high costs of donating.
BIO: Tinglong Dai is an assistant professor at the Johns Hopkins University, Carey Business School. His research areas include healthcare operations, and marketing/operations interfaces. He received his PhD (2013) and MS (2009) in Operations Management/Robotics from Tepper School of Business, Carnegie Mellon University. He also received an MPhil (2006) in Industrial Engineering and Engineering Management from the Hong Kong University of Science and Technology.
Dai’s research has been published in leading journals such as Management Science, Operations Research, Manufacturing & Service Operations Management, and INFORMS Journal on Computing. He is on the Editorial Review Board of Production and Operations Management. He is the founder of the Johns Hopkins Symposium on Healthcare Operations, and co-edits the Handbook of Healthcare Analytics: Theoretical Minimum for Conducting 21st Century Research on Healthcare Operations, to be published by John Wiley & Sons in 2018.
Dai is the recipient of the inaugural Johns Hopkins Discovery Award (2015), Dean’s Award for Faculty Excellence (2016, 2017), M&SOM Meritorious Service Award (2016), the First Place of POMS Best Healthcare Paper Award (2012), INFORMS Pierskalla Runner-up Award for the Best Paper in Healthcare (2012), and the Second Place of POMS Best Healthcare Paper Award (2016). He is a finalist for the Elwood S. Buffa Doctoral Dissertation Award (2014) and the POMS College of Supply Chain Management Best Student Paper Award (2013). He has been quoted in Washington Post, Baltimore Sun, MedPage Today, and Pharmacy Times, among others.
]]>
ABSTRACT:
A well-studied deterministic algorithmic technique for convex optimization is the class of so-called Interior Point Methods of Nesterov and Nemirovski, which involve taking a sequence of Newton steps along the "central path" towards the optimum. An alternative randomized method, known as simulated annealing, involves performing a random walk around the set while "cooling" the stationary distribution towards the optimum. We will show that these two methods are, in a certain sense, fully equivalent: both techniques can be viewed as different types of path following. This equivalence allows us to get an improved state-of-the-art rate for simulated annealing, and provides a new understanding of random walk methods using barrier functions.
BIO: Jacob Abernethy is Assistant Professor in Computer Science at Georgia Tech. He started his faculty career in the Department of Electrical Engineering and Computer Science at the University of Michigan. In October 2011 he finished a PhD in the Division of Computer Science at the University of California at Berkeley, and then spent nearly two years as a Simons postdoctoral fellow at the CIS department at UPenn, working with Michael Kearns. Abernethy's primary interest is in Machine Learning, with a particular focus in sequential decision making, online learning, online algorithms and adversarial learning models. He did his Master's degree at TTI-C, and his Bachelor's Degree at MIT. Abernethy's PhD advisor is Prof. Peter Bartlett
]]>We strongly encourage students to act now to seek full-time employment, internships, and projects (rather than waiting until the end of the semester). Plus, enjoy a free pizza lunch!
Where: ISyE Main Bldg, 2nd Floor Atrium
When: Thursday, September 7 | 11:00am - 2:30pm
What: The session will include:
Please plan on staying for the duration of the event and bring copies of your resume and business cards. Dress is business casual.
EVENT SPONSOR
The event is sponsored through the generosity and support of JP Morgan Chase & Co. and APICS - Atlanta Chapter. APICS is a non-profit educational organization addressing operations management and supply chain management issues, and providing professional development opportunities to its members. APICS Membership is free for full time students. Join today at www.apics.org/join and start networking at local APIC Atlanta events. Also make sure to stop by the APICS table at the event.
]]>]]>
ABSTRACT:
In monopoly pricing situations, firms should optimally vary prices to learn demand. The variation must be sufficiently high to ensure complete learning. In competitive situations, however, varying prices provides information to competitors and may reduce the value of learning. Such situations may arise in the pricing of new products such as pharmaceuticals. This talk will discuss how this effect can be strong enough to stop learning so that firms optimally reduce any variation in prices and choose not to learn demand. The result can be that the selling firms achieve a collaborative outcome instead of a competitive equilibrium. The result has implications for policies that restrict price changes or require disclosures.
Bio: John R. Birge is the Jerry W. and Carol Lee Levin Distinguished Service Professor of Operations Management at the University of Chicago Booth School of Business. Previously, he was Dean of the McCormick School of Engineering and Applied Science and Professor of Industrial Engineering and Management Sciences at Northwestern University. He also served as Professor and Chair of Industrial and Operations Engineering at the University of Michigan, where he also established the Financial Engineering Program. He is former Editor-in-Chief of Mathematical Programming, Series B and former President of INFORMS. His honors and awards include the IIE Medallion Award, the INFORMS Fellows Award, the MSOM Society Distinguished Fellow Award, the Harold W. Kuhn Prize, the George E. Kimball Medal, the William Pierskalla Award, and election to the US National Academy of Engineering. He received M.S. and Ph.D. degrees from Stanford University in Operations Research, and an A.B. in Mathematics from Princeton University.
]]>This year’s event will focus on Industry Disruptors. Registration and networking will begin at 10:30 a.m. with the luncheon and program scheduled to begin promptly at 11:30 a.m. and conclude by 1:00 p.m.
The event will feature speakers from Georgia Department of Economic Development, HunterMaclean, UPS, Manhattan Associates and the Supply Chain & Logistics Institute at Georgia Tech.
Registration is $25.00 and more information and online registration is available at www.georgia.org/LogisticsLunch.
]]>Complex supply chain transformation requires managing resources from many different departments, ensuring internal and external stakeholder alignment, mitigating large amounts of risk, and implementing communication, risk mitigation, and change management plans to ensure a successful project. Successful project management in complex supply chain environments requires application of well-planned integrated approaches. This course conveys an integrated view to supply chain transformation incorporating elements of change management, test plan development, project management techniques, and establishing effective project management teams.
This course provides a deeper understanding of the PMBOK knowledge areas of project integration and procurement applied in the supply chain vendor selection and management process. To keep pace with the continuous moves toward outsourcing of operations and the advancement of technology, companies need to focus on selecting the right suppliers and partnerships to provide the most value to their customers and to remain profitable. This course provides the knowledge, skills, and tools to ensure that you are selecting the right supply chain partners (including 3PL’s) based on your business goals. Emphasis is placed on understanding alternative techniques for supplier selection including applied quantitative decision making techniques.
With your RSVP, please let us know what GT program you were in. We look forward to seeing you!
Please RSVP at http://evite.me/UsYUDPKfxA
Go Jackets!
]]>ABSTRACT:
This research extends and simplifies existing piecewise-linear homotopy (PL) methods to solve G(x) = 0, with G : Rn → Rm. Existing PL methods are designed to solve F (x) = 0, with F : Rn → Rn and some related point-to-set mappings. PL methods are a component of what is also known as numerical continuation methods, and they are known for being globally convergent methods. First, we present a new PL method for computing zeros of functions of the form f : Rn → R by mimicking classical PL methods for computing zeros of functions of the form f : R → R. Our PL method avoids traversing subdivisions of Rn × [0, 1] and instead uses an object that we refer to as triangulation-graph, which is essentially a triangulation of R × [0, 1] with hypercubes of Rn as its vertices. The hypercubes are generated randomly, and a sojourn time of an associated discrete-time Markov chain is used to show that not too many cubes are generated. Thereafter, our PL method is applied to solving G(x) = 0 for G : Rn → Rm under inequality constraints. The resultant method for solving G(x) = 0 translates into a new type of iterative method for solving systems of linear equations. Some computational illustrations are reported. A possible application to optimization problems is also indicated as a direction for further work.
Bio: Ira Monroe Wheaton Jr. was born and raised in Chicago, IL to parents Ira Sr. and Laura. From an early age, Ira excelled in school and discovered a great affinity for learning. After graduating as the valedictorian of his high school class, Ira decided to pursue a Bachelor of Science degree in Mathematics. In 2007, Florida A&M University (FAMU) provided Ira with a full academic scholarship to accomplish this goal. During his time at FAMU, Ira was active on campus as a tutor in the FAMU Math Lab and a member of many organizations including Florida-Georgia Louis Stokes Alliance for Minority Participation (FGLSAMP), Kemetic Mathematical Society, and Honors Student Association. Furthermore, he participated in summer internships at the NASA Johnson Space Center in Houston, TX and The Northern Trust Company in Chicago, IL.
After obtaining his B.S. degree with Summa Cum Laude honors in April 2011, Ira enrolled in the Florida State University (FSU) Department of Mathematics and obtained an M.S. in Financial Mathematics in May 2013. Thereafter, Ira enrolled in the FSU Department of Industrial and Manufacturing Engineering (IME) to pursue a PhD. During that time, he served as a teaching assistant, working closely with professors to assist them with Operations Research I & II courses. With the support of the Florida Education Fund’s McKnight Doctoral Fellowship, Ira has been involved in research on homotopy methods, which has resulted in a publication in the Journal of Computational and Applied Mathematics, as well as three more articles in progress. Ira graduated with a PhD in Industrial Engineering on May 6, 2017.
Ira is currently serving as an Instructor in the Morehouse College Department of Mathematics. In his spare time, Ira enjoys spending time with his wife and daughter, playing the piano, and playing basketball. He hopes to continue to contribute to the field and to inspire new researchers in the same way that his advisor and others have done for him.
]]>ABSTRACT:
The presence of symmetries in integer programs is well known to hurt the performance of branch-and-cut methods and several symmetry handling methods have been proposed. This talk will give an overview on these methods and investigate their computational impact. Most of these techniques perform pruning in the tree or fixing variables. As an alternative, a general polyhedral approach will be presented that is based on the convex hull of lexicographically maximal points within their orbit, so-called symretopes. While a complete description of these polytopes is only known for special symmetry groups, one can use their structure to construct efficiently solvable integer programming formulations. Computational results will show that this approach is competitive with the state-of-the-art methods based on pruning the tree.
]]>ABSTRACT:
Sparsity plays a key role in linear statistical modeling and beyond. In this talk I will discuss the best subset selection problem, a central problem in statistics, wherein the task is to select a set of k relevant features from among p variables, given n samples. I will discuss recent computational techniques relying on integer optimization and first order optimization methods, that enable us to obtain high-quality, near-optimal solutions for best-subsets regression, for sizes well beyond what was considered possible. This sheds interesting new insights into the statistical behavior of subset selection problems vis-a-vis popular, computationally friendlier methods like L1 regularization -- thereby motivating the design of new statistical estimators with better statistical and computational properties. If time permits, I will also discuss another closely related, extremely effective, but relatively less understood sparse regularization scheme: the forward stage-wise regression (aka Boosting) in linear models.
]]>ABSTRACT:
The National Science Foundation offers a number of funding opportunities for investigators working in the field of operations research, both within the disciplinary programs in Engineering and other directorates, and through cross-cutting initiatives that are foundation-wide. This presentation will describe opportunities that are relevant to the Industrial and Operations Engineering communities, with particular emphasis on the Operations Engineering program in the Division of Civil, Mechanical, and Manufacturing Innovation. For those not familiar with NSF, opportunities for junior investigators and an introduction to the merit review process will also be described.
BIO: Georgia-Ann Klutke is the Program Director for the Operations Engineering program at the National Science Foundation. She retired in 2015 as Professor in the Industrial and Systems Department at Texas A&M University. She has also served on the faculties of The University of Texas at Austin and the University of Massachusetts, Amherst. She holds a B.S. (Mathematics) and M.S. (Biostatistics) from the University of Michigan and a Ph.D. (Industrial Engineering and Operations Research) from VPI&SU. She is a Fellow of the Institute of Industrial and Systems Engineers.
]]>ABSTRACT:
In many applications, we have access to large datasets (such as location of major road intersections in a state, healthcare records, database of building profiles, and visual stimuli), but are limited in how many labels (such as traffic or wind speed at the intersections, customer satisfaction, energy usage, and brain response, respectively) can be collected under budget constraints. Classical experimental design in statistics addresses this problem, however the solutions tend to be combinatorial. We consider computationally tractable methods for the experimental design problem, where k out of n design points of dimension p are selected so that certain optimality criteria are approximately satisfied. We prove a constant approximation ratio under a very weak condition that k > 2p, and a (1 + eps) relative approximation ratio under slightly stronger conditions in which k is still a linear function of p. Our results are based on a convex relaxation of the combinatorial problem, followed by different sampling strategies. We also present numerical results on both synthetic and real-world design problems that verify the practical effectiveness of the proposed algorithm.
BIO: Aarti Singh received her B.E. in Electronics and Communication Engineering from the University of Delhi in 2001, and M.S. and Ph.D. degrees in Electrical and Computer Engineering from the University of Wisconsin-Madison in 2003 and 2008, respectively. She was a Postdoctoral Research Associate at the Program in Applied and Computational Mathematics at Princeton University from 2008 to 2009, before joining the School of Computer Science at Carnegie Mellon in 2009 where she is currently an Associate Professor. Her research interests lie at the intersection of machine learning, statistics and signal processing, and focus on designing statistically and computationally efficient algorithms that can leverage inherent structure of the data in the form of clusters, graphs, subspaces and manifold using direct, compressive and active queries. Her work is recognized by an NSF Career Award, a United States Air Force Young Investigator Award, A. Nico Habermann Faculty Chair Award, Harold A. Peterson Best Dissertation Award, and a best student paper award at Allerton.
]]>ABSTRACT:
Stochastic Gradient Langevin Dynamics (SGLD) is a popular variant of Stochastic Gradient Descent, where properly scaled isotropic Gaussian noise is added to an unbiased estimate of the gradient at each iteration. This modest change allows SGLD to escape local minima and suffices to guarantee asymptotic convergence to global minimizers for sufficiently regular non-convex objectives. I will present a nonasymptotic analysis in the context of non-convex learning problems and show that SGLD requires $O(\epsilon^{-4})$ iterations to sample $O(\epsilon)$-approximate minimizers of both empirical and population risk, where $\tO(\cdot)$ hides polynomial dependence on a temperature parameter, the model dimension, and a certain spectral gap parameter. As in the asymptotic setting, the analysis relates the discrete-time SGLD Markov chain to a continuous-time diffusion process. A new tool that drives the results is the use of weighted transportation cost inequalities to quantify the rate of convergence of SGLD to a stationary distribution in the Euclidean $2$-Wasserstein distance. This talk is based on joint work with Sasha Rakhlin and Matus Telgarsky.
Bio: Maxim Raginsky received the B.S. and M.S. degrees in 2000 and the Ph.D. degree in 2002 from Northwestern University, all in electrical engineering. He has held research positions with Northwestern, University of Illinois at Urbana-Champaign (where he was a Beckman Foundation Fellow from 2004 to 2007), and Duke University. In 2012, he returned to UIUC, where he is currently an Assistant Professor and William L. Everitt Fellow in Electrical and Computer Engineering. He is also a faculty member of the Coordinated Science Laboratory. Dr. Raginsky received a Faculty Early Career Development (CAREER) Award from the National Science Foundation in 2013. His research interests lie at the intersection of information theory, machine learning, and control. He is a member of the editorial boards of Foundations and Trends in Communications and Information Theory and IEEE Transactions on Network Science and Engineering.
]]>ABSTRACT:
For time series with long memory, inference through resampling is of particular importance, since the asymptotic distributions are often difficult to determine statistically. To establish the asymptotic validity of certain resampling procedures, it requires a fine understanding on the dependence between two finite blocks of the time series. We shall introduce some recent progress on this direction.
]]>
ABSTRACT:
Markets for wholesale electricity supply are now ubiquitous throughout the industrialized world. In the simplest form of these markets, a perfectly competitive partial equilibrium corresponds to the optimal solution to a convex economic dispatch problem, where Lagrange multipliers yield nodal energy prices. This construction can be extended to the setting where the dispatch problem becomes a convex stochastic program (for example to deal with intermittent renewable energy, or uncertain hydro-reservoir inflows) and agents maximize expected profits. When agents are risk averse, competitive partial equilibrium corresponds to a risk-adjusted social optimum as long as derivative instruments enable agents to trade risk. We illustrate these ideas using some simple examples drawn from the New Zealand electricity market.
Bio: Dr. Andy Philpott is a Professor at University of Auckland, and he is the director of the Electric Power Optimization Centre. He received his PhD and MPhil from University of Cambridge, and his BSc and BA degrees from Victoria University of Wellington. His research interests span most of mathematical programming, in particular linear, non-linear and stochastic programming and their application to operations research problems, in particular optimal planning under uncertainty, capacity expansion in telecommunications and power networks, optimal power generation hydro-electric power systems, stochastic optimization in supply chains, and optimal yacht routing under uncertainty. Much of his recent research has been conducted as part of the Electric Power Optimization Centre, which develops optimization and equilibrium models of electricity markets.
]]>ABSTRACT:
We characterize the streaming space complexity of every symmetric norm l (a norm on R^n invariant under sign-flips and coordinate-permutations), by relating this space complexity to the measure-concentration characteristics of l. Specifically, we provide nearly matching upper and lower bounds on the space complexity of calculating a (1±ϵ)-approximation to the norm of the stream, for every 0<ϵ≤1/2. (The bounds match up to poly(ϵ^−1logn) factors.) We further extend those bounds to any large approximation ratio D≥1.1, showing that the decrease in space complexity is proportional to D^2, and that this factor the best possible. All of the bounds depend on the median of l(x) when x is drawn uniformly from the l2 unit sphere. The same median governs many phenomena in high-dimensional spaces, such as large-deviation bounds and the critical dimension in Dvoretzky's Theorem.
The family of symmetric norms contains several well-studied norms, such as all l_p norms, and indeed we provide a new explanation for the disparity in space complexity between p≤2 and p>2. In addition, we apply our general results to easily derive bounds for several norms that were not studied before in the streaming model, including the top-k norm and the k-support norm, which was recently employed for machine learning tasks. Overall, these results make progress on two outstanding problems in the area of sublinear algorithms (Problems 5 and 30 in https://sublinear.info). Based on paper: https://arxiv.org/abs/1511.01111. Joint work with: Jaroslaw Blasiok, Vladimir Braverman, Stephen R. Chestnut, Robert Krauthgamer.
Bio: Lin Yang is currently a PhD student in the Computer Science Department and Physics Department of Johns Hopkins University. He will obtain two PhD degrees at the end of this year. He works on algorithms for streaming data analysis, studying the fundamental complexity of the streaming computation model as well as designing new and better algorithms for important problems. His research connects theoretical computer science with machine learning and computational cosmology in the big data regime.
]]>ABSTRACT:
We consider a set of collaborative agents that need to coordinate their actions so as to socially optimize the sum of their objectives while satisfying a common resource constraint. The objective functions of the players are unknown to them a priori and are revealed in an online manner over time. Given a resource allocation, each player’s action is determined by solving an integer program. Due to privacy issues, players want to share limited information while solving this problem in a decentralized way. A cardinality resource constraint links all player actions. The resulting problem is an online optimization problem to optimally allocate the resource among the players prior to observing the item values. The performance of an online algorithm is measured by regret, defined as the difference between the total gain of the best-fixed decision considering all data in hindsight and the total gain incurred by the online decisions we make. A good online algorithm is expected to have regret as a sublinear function of the number of rounds, T. In this research, we show that for any deterministic online algorithm for the resource allocation problem, there exist objective function coefficients that guarantee O(T) regret. Furthermore, for the case when the players' integer programs satisfy a special concavity condition, we propose a randomized online algorithm for the resource allocation problem that guarantees an upper bound of O(\sqrt T) on the expected regret.
]]>
The efficient flow of inbound materials through your supply chain can have a significant impact on cost, quality, service, and delivery to your customer. With continued pressures from globalization, market volatility, and innovation expectations, companies are extending their reliance on lean principles to outside the four walls of the manufacturing facility. Lean inbound logistics requires strategic coordination to synchronize suppliers, transportation providers and multiple supply chain partners. And like all complex multi-party operations, the road to improvement will have its challenges. Such hurdles could include balancing inventory cost with transportation cost, the lack of leadership understanding and support, and the realities of incumbent manufacturing and procurement contracts and systems.
This course focuses on the why and how to implement lean logistics to support the lean supply chain operations – both in manufacturing and distribution. Areas of focus include: lean strategy development, identifying and overcoming operational realities, and learning the tools necessary for successful implementation.
Special Note: Includes a tour of the port of Savannah
Company owners, consultants, logistics service providers, chief supply chain officers, vice presidents of sales operations, vice presidents and directors of process improvement, and executive, senior, vice presidents, and directors of supply chain, logistics, procurement, manufacturing, or distribution
Complex supply chain transformation requires managing resources from many different departments, ensuring internal and external stakeholder alignment, mitigating large amounts of risk, and implementing communication, risk mitigation, and change management plans to ensure a successful project. Successful project management in complex supply chain environments requires application of well-planned integrated approaches. This course conveys an integrated view to supply chain transformation incorporating elements of change management, test plan development, project management techniques, and establishing effective project management teams.
IPIC 2017 aims to provide an open forum for researchers, innovators and practitioners to introduce leading edge concepts and methodologies; to review the state-of-the-art technologies and latest projects, and to identify critical issues and challenges for future Physical Internet induced research, innovation and implementation. For 2017, the comprehensive view on the Physical Internet from the IPIC conference merges with the technical and intralogistic view from the Logistikwerkstatt Graz for a promising format. Both the IPIC and the Logistikwerkstatt Graz are well established international conferences with broad contribution from industry and research.
Visit www.pi.events to learn more about the conference.
Call for Papers and Contributions
IPIC 2017 invites you to submit an original research or innovation contribution to the conference.
Contributions can take three forms: papers, posters and presentations. Papers and posters will be made widely available on www.pi.events and physicalinternetinitiative.org. Visit www.pi.events/call-for-papers for guidelines, templates, and more information relating to paper submission.
]]>
.
]]>ABSTRACT:
It is common practice among the international organizations to resort the local transport companies for delivering food aid in regions plagued by hunger and malnutrition. The reliability of the transportation services provided by these third parties is essential for ensuring the food security, much needed in these regions. The transporters in developing countries often inflate their tariffs in order to protect themselves from the expected price increases. They also allocate their trucks to secure jobs in the spot market when the contract rates are significantly lower than that of the spot market, which jeopardizes service quality. Using data gathered from the Kenya-based operations of the United Nation’s World Food Programme, we demonstrate that the randomness in transport spot market prices can be attributed to the fuel prices. Therefore, we devise a new barrier-type options contract that can be adopted for reducing the transporter’s downside risk and mitigating their incentive to bid defensively. Our findings through the numerical experiments on the real-life data set are encouraging since they demonstrate a noteworthy potential that both the food aid organization and the transporters can benefit from the implementation of the proposed contract scheme.
Short Bio: Vedat Verter is James McGill Professor of Operations Management at the Desautels Faculty of Management of McGill University. He specializes on the application of business analytics for policy design and decision-making in the public sector. His areas of research are transport risk management, sustainable operations and healthcare operations management. His work in these areas is well recognized through top tier journal publications as well as invited presentations around the globe. In the area of healthcare, he focuses on preventive, primary, emergency, acute and chronic care processes, as well as their interaction. He is Founding Director of the NSERC CREATE Program in Healthcare Operations and Information Management, a seven-University PhD/PDF training program across Canada. Professor Verter is also Editor-in-Chief of Socio-Economic Planning Sciences, an international journal focusing on public sector decision-making.
]]>
Abstract: A typical optimization model in operations research allocates limited resources among competing activities to derive an optimal portfolio of activities. In contrast, practitioners often form a rank-ordered list of activities, and select those with the highest priority, at least when choosing an activity is a yes-no decision. Ranking schemes that score activities individually are well known to be inferior. So, we describe a class of two-stage stochastic integer programs that accounts for structural and stochastic dependencies across activities and constructs an optimized priority list. We further discuss a class of optimization models, subject to a single "budget" constraint, that naturally leads to a family of optimal nested solutions at certain budget increments. We use several applications to both motivate the work and illustrate results, ranging from a stochastic facility location model to a hierarchical graph clustering problem. We also describe possible extensions.
Bio: David Morton is a Professor of Industrial Engineering and Management Sciences at Northwestern University. His research interests include stochastic and large-scale optimization with applications in security, public health, and energy systems. He received a B.S. in Mathematics and Physics from Stetson University and an M.S. and Ph.D. in Operations Research from Stanford University. Prior to joining Northwestern, he was on the faculty at the University of Texas at Austin, worked as a Fulbright Research Scholar at Charles University in Prague, and was
]]>ABSTRACT:
We discuss the framework of Blind Regression (also known as Latent Variable Model) motivated by the problem of Matrix Completion for recommendation systems: given n users and m movies, the goal is to predict the unknown rating of a user for a movie using known observations, i.e. completing the partially observed matrix. We posit that each user and movie is associated with latent features, and the rating of a user for a movie equals the noisy version of latent function applied to the associated latent features. The goal is to predict such a function value for user-movie pairs for which ratings are unknown, just like the classical regression setting. However, unlike the setting of regression, features are not observed here — hence the term Blind Regression. Such a model arises as a canonical characterization due to multi-dimensional exchangeability property a la Aldous and Hoover (early 1980s).
In this talk, using inspiration from the classical Taylor’s expansion for differentiable functions, we shall propose a prediction algorithm that is consistent for all Lipschitz continuous functions. We provide finite sample analysis that suggests that even when observing a vanishing fraction of the matrix, the algorithm produces accurate predictions. We discuss relationship with spectral algorithm for matrix completion, and the collaborative filtering.
The talk is based on joint works with Christina Lee, Yihua Li and Dogyoon Song (MIT).
BIO: Devavrat Shah is a Professor with the department of Electrical Engineering and Computer Science at Massachusetts Institute of Technology. His current research interests are at the interface of Statistical Inference and Social Data Processing. His work has been recognized through prize paper awards in Machine Learning, Operations Research and Computer Science, as well as career prizes including 2010 Erlang prize from the INFORMS Applied Probability Society and 2008 ACM Sigmetrics Rising Star Award. He is a distinguished young alumni of his alma mater IIT Bombay.
]]>ABSTRACT:
This talk gives a general overview of practical computational methods for computing branch decompositions of matroids and their usage for problems related to matroids such as solving integer programs and others. The concept of branch decompositions and its related invariant branchwidth were first introduced by Robertson and Seymour in their proof of the Graph Minors Theorem and can be easily generalized for any symmetric submodular set function. This talk is based on joint work with John Arellano, Edray Goins, Jing Ma, Susan Margulies, Nolan McMurray.
Bio:
Illya V. Hicks was born and raised in Waco, TX. He received a BS in mathematics (1995) from Southwest Texas State University (currently Texas State University at San Marcos). He also received an MA and PhD in Computational and Applied Mathematics (2000) from Rice University. Illya served as faculty member in the Industrial and Systems Engineering Department at Texas A&M University (2000-2006) and is currently a professor in the Computational and Applied Mathematics Department at Rice University where he also serves as faculty advisor to the president of Rice University.
In terms of research, his interests are in combinatorial optimization, graph theory, and integer programming with applications big data, imaging, social networks, and logistics. Illya is also the recipient of the 2005 Optimization Prize for Young Researchers from the Optimization Society of the Institute for Operations Research and the Management Sciences (INFORMS) and the 2010 Forum Moving Spirit Award from INFORMS for his work with the Minority Issues Forum of INFORMS.
]]>ABSTRACT:
Driven by the development of sensing, communication, and cloud-based data management platform, smart tele-service systems have seen tremendous growth in recent years. The smart tele-service system can monitor the operation of a product/system at real time and provide service alerts to the end-users when an imminent failure is predicted. The unprecedented data availability in smart tele-service systems provides significant opportunities for the data-drive failure prognosis but, at the same time, reveals some critical challenges. First, the heterogeneous data with diverse data types often hinder to establish a unified prognostic framework. Second, individual-level data has become available in large scale and consequently, there is a pressing need for individualized modeling and prognosis. Lastly, severe signal noise and non-stationary behavior in the monitoring data need to be addressed appropriately. To address those challenges, a series of data-driven prognostic methodologies have been proposed by my research group. First, a flexible hierarchical Bayesian model namely the joint prognostic model framework is proposed. The joint prognostic model integrates information from diverse data types and considers the data heterogeneity through the mixed-effects model. Leveraging on Bayesian theory, our prognostic model provides highly individualized failure prediction based on the smart tele-service system. Furthermore, the joint prognostic model has been extended to address the severe noise issue and the non-stationary behavior in the monitoring data. Those extensions increase the accuracy of the failure prediction significantly. All methods have shown advantageous features in both numerical studies and case studies with real world data from the smart tele-service systems in the application of automotive battery failure prognosis
Bio: Shiyu Zhou is a Professor in the Department of Industrial and Systems Engineering at the University of Wisconsin-Madison. He received his B.S. and M.S. in Mechanical Engineering from the University of Science and Technology of China in 1993 and 1996, respectively, and his master’s in Industrial Engineering and Ph.D. in Mechanical Engineering from the University of Michigan in 2000. His research interests include industrial data analytics for quality and productivity improvement methodologies by integrating statistics, system and control theory, and engineering knowledge. He is a recipient of a CAREER Award from the National Science Foundation and the Best Application Paper Award from IIE Transactions. He is currently the editor for the Design and Manufacturing focus issue of IISE Transactions, a fellow of ASME and member of IIE, INFORMS, and SME.
]]>ABSTRACT:
Testing the population mean is fundamental in statistical inference. When the dimensionality of a population is high, traditional Hotelling's T2 test becomes practically infeasible. In this paper, we propose a new testing method for high-dimensional mean vectors. The new method projects the original sample to a lower-dimensional space and carries out a test with the projected sample. We derive the theoretical optimal direction with which the projection test possesses the best power under alternatives. We further propose an estimation procedure for the optimal direction, so that the resulting test is an exact $t$-test under the normality assumption and an asymptotic chi-square test with 1 degree of freedom without the normality assumption. Monte Carlo simulation studies show that the new test can be much more powerful than the existing methods, while it also well retains Type I error rate.
]]>
ABSTRACT:
Smart grid involves the imposition of an advanced cyber layer atop the physical layer of the electricity grid, in order to improve the efficiency, security and cost of electricity use and distribution, and to allow for greater decentralization of power generation and management. This cyber-physical setting motivates a number of problems in network analysis, and this talk will briefly describe several of these problems together with approaches to solving them. These include competitive privacy in which multiple grid entities seek an optimal trade-off between privacy lost and utility gained from information sharing; distributed inference in which both the cyber and physical network topologies have roles to play in achieving consensus; real-time topology identification which helps in the mitigation of cascading failures; and attack construction which seeks an understanding of optimal strategies for attacking the grid in support of the design of effective countermeasures.
BIO: H. Vincent Poor is the Michael Henry Strater University Professor of Electrical Engineering at Princeton University. His research interests are in the areas of information theory and signal processing, and their applications in wireless networks, smart grid and related fields. Dr. Poor is a member of the National Academy of Engineering and National Academy of Sciences, and a foreign member of the Royal Society. Recent recognition of his work includes the 2016 John Fritz Medal, the 2017 IEEE Alexander Graham Bell Medal, and honorary doctorates and professorships from several universities.
]]>
ABSTRACT:
In recent years, personal location data is continuously captured by mobile devices, GPS chips and other sensors. Such data provides a unique learning opportunity on individuals’ mobility behavior that may be used for various applications in transportation, marketing, homeland security and smart cities. Nonetheless, modeling such data poses new challenges related to data volume, diversity, inhomogeneity and the required granularity level. In this talk, we will address a real ‘smart city’ use-case and cover some of its associated opportunities and challenges. We will present a new set of mobility-behavior models that generalizes Markov Chains and Variable-Order Bayesian Networks. We will discuss how they can be used in different smart city applications such as pattern recognition, anomaly detection, clustering and classification.
Bio:
Irad Ben-Gal is a visiting professor at MS&E Stanford University and a full professor in the Department of Industrial Engineering at Tel Aviv University, heading the AI & Business Analytics lab. His research focuses on applied probability, machine learning and information theory applications to industrial and service systems. He wrote 3 books, published more than 80 scientific papers and patents and received several best papers awards. He is a Department Editor in IIE Transactions and serves on the Editorial Boards of several data science journals. Irad led various R&D projects and worked with companies such as Siemens, Intel, Applied Materials, GM, Nokia, AT&T, Oracle and SAP. Irad is the co-founder and active chairman of CB4 (“See Before”), a startup backed by Sequoia Capital that provides granular predictive analytics solutions to retail organizations.
ABSTRACT:
In this talk, we review new theory that explains the effectiveness of sparse flexibility design in any finite size manufacturing network. Under stochastic demand, we establish two fundamental properties of sparse designs, a supermodularity property and a decomposition property. These properties are then used to provide the first theoretical justification for several well-known observations in the process flexibility literature, and to derive new insights into designing flexible processes in large systems. Under worst-case demand, we propose the plant cover index and establish its relation with the worst-case sales. Applying this relation, we demonstrate the effectiveness of a certain sparse design, called the long chain. Finally, we discuss the combination of process flexibility and strategic inventory as an effective disruption mitigation strategy.
BIO:
David Simchi-Levi is a Professor of Engineering Systems at MIT and Chairman of Opalytics, a cloud analytics platform company. He is considered one of the premier thought leaders in supply chain management and business analytics. His research focuses on developing and implementing robust and efficient techniques for operations management. He has published widely in professional journals on both practical and theoretical aspects of supply chain and revenue management.
His Ph.D. students have accepted faculty positions in leading academic institutes including U. of California Berkeley, Columbia U., Cornell U., Duke U., Georgia Tech, Harvard U., U. of Illinois Urbana-Champaign, U. of Michigan, Purdue U. and Virginia Tech.
Professor Simchi-Levi co-authored the books Managing the Supply Chain (McGraw-Hill, 2004), the award winning Designing and Managing the Supply Chain (McGraw-Hill, 2007) and The Logic of Logistics (3rd edition, Springer 2013). He also published Operations Rules: Delivering Customer Value through Flexible Operations (MIT Press, 2011).
He served as the Editor-in-Chief for Operations Research (2006-2012), the flagship journal of INFORMS and for Naval Research Logistics (2003-2005). He is an INFORMS Fellow, MSOM Distinguished Fellow and the recipient of the 2014 INFORMS Daniel H. Wagner Prize for Excellence in Operations Research Practice; 2014 INFORMS Revenue Management and Pricing Section Practice Award; 2009 INFORMS Revenue Management and Pricing Section Prize and Ford 2015 Engineering Excellence Award.
Professor Simchi-Levi has consulted and collaborated extensively with private and public organizations. He was the founder of LogicTools which provided software solutions and professional services for supply chain optimization. LogicTools became part of IBM in 2009. In 2012 he co-founded OPS Rules, an operations analytics consulting company. The company became part of Accenture in 2016.
]]>ABSTRACT:
Inspired by communication and healthcare services, this talk summarizes the methods developed with many collaborators over the decades to understand the transient behavior of dynamic rate queues. This analysis is needed when confronted with the dynamic parameters found in time-inhomogeneous Markovian queueing models. The static equilibrium analysis for the steady state of constant rate queues no longer applies.
Constant parameters summarizing the transient behavior for these steady state systems are now replaced by deterministic dynamical systems. We can then approximate the optimal behavior of these queues by controlling the corresponding family of ordinary differential equations.
Bio:
William A. Massey is the Edwin S. Wilsey Professor in the Department of Operations Research and Financial Engineering at Princeton University. He is both an inaugural fellow of the American Mathematical Society (AMS) and a fellow of the Institute for Operations Research and the Management Sciences (INFORMS). His research interests include dynamic rate queues, queueing networks, stochastic analysis, dynamical systems control, and their applications to communications and healthcare management. He received his A.B. degree in mathematics from Princeton University in 1977 (Magna Cum Laude, Phi Beta Kappa, and Sigma Xi) and his Ph.D. in mathematics from Stanford University in 1981. The latter was funded by the Bell Laboratories Cooperative Research Fellowship for Minorities. He worked at Bell Laboratories as a member of technical staff in their Mathematical Sciences Research Center for 20 years until joining the Princeton University faculty in 2001.
]]>OVERVIEW
Building a Plan for Every Part (PFEP) is one of the most effective ways to create standardization in an ordering and delivery process and optimize material flow. Breakthrough results in on-time-delivery, coverage, and cost reduction can be achieved through this discipline, but how do you do it?
In this 1-hour live webinar, join Instructor Brad Bossence as we discuss the critical components of PFEP and explore:
WHO SHOULD ATTEND
Leaders in manufacturing, logistics, supply chain, quality, transportation, materials, warehousing, administration, and other functional areas responsible for moving material and/or information.
ABOUT THE PRESENTER
Brad Bossence is instructor of the Georgia Tech "Introduction to Plan for Every Part (PFEP) and Inventory Sizing" course and Vice President of LeanCor Supply Chain Group. He also teaches other Georgia Tech short courses including Lean Warehousing.
]]>We strongly encourage students to act now to seek full-time employment, internships, and projects (rather than waiting until the end of the semester). Plus, enjoy a free pizza lunch!
Where: ISyE Main Bldg, 2nd Floor Atrium
When: Tuesday, March 14th | 11:15am - 2:15pm
What: The session will include:
Please plan on staying for the duration of the event and bring copies of your resume and business cards. Dress is business casual.
EVENT SPONSOR
The event is sponsored through the generosity and support of APICS - Atlanta Chapter. APICS is a non-profit educational organization addressing operations management and supply chain management issues, and providing professional development opportunities to its members. APICS Membership is free for full time students. Join today at www.apics.org/join and start networking at local APIC Atlanta events. Also make sure to stop by the APICS table at the event.
]]>]]>
ABSTRACT:
Worldwide, cancer is a leading cause of death, and metastatic breast cancer (MBC) is the top cause of cancer death among women. We present a data-driven approach to planning clinical trials and designing novel drug therapies for metastatic breast cancer. First, we describe construction of a large database of MBC clinical trial results and tools to help clinicians visualize the data. Next, we use statistical models to predict efficacy and toxicity outcomes of trials before they are run, with implications for selecting between multiple drug therapies for testing. Finally, we use optimization models to design novel therapies that strike a balance between improving patient outcomes and learning about new drugs; we present evidence that these models may improve trial outcomes compared to current practice.
BIO
John Silberholz is currently a Lecturer at MIT Sloan. He received his PhD in 2015 from the MIT Operations Research Center, and a BS in computer science and a BS in mathematics from the University of Maryland. John's research interests lie in healthcare analytics - broadly construed - with a current focus on applications to cancer screening and treatment. John is a recipient of the William Pierskalla Best Paper Award (2013) for the top healthcare management science paper, the INFORMS Undergraduate Operations Research Prize, and an NSF Graduate Research Fellowship.
]]>A limited number of "All Access" passes will be made available to students in exchange for a day of blogging (grants access to activities in addition to the expo…stop by the SCL suite - Groseclose 228 or email event@scl.gatech.edu for more information).
Home Delivery World brings decision-making retailers together to discuss strategies and learn from their peers. The event will deliver 5 streams of unique content for retailers, etailers, grocers, logistics, warehousing and solution providers looking to discuss the latest delivery operations challenges and trends. For its fifth anniversary, HDW2017 is bringing together etailers, retailers, service providers, and for the first time, grocers from The Home Depot, Boxed, IBM, Birchbox, XPO Logistics, Haverty’s, Overstock.com, Tiffany & Co., Descartes, Verizon, Lowe’s Foods, Giant Eagle, MXD Group, Unilever and many more to discuss regional, domestic, and international strategies. It wouldn’t be Home Delivery World without unparalleled networking and learning opportunities that ensure you leave with the knowledge and connections that you need to finish in first place.
Some key topics at the conference will be:
If you are interested in home delivery or discussing topics like reverse logistics, direct fulfillment, omnichannel fulfillment, multichannel returns, real-time visibility for inventory, and international expansion you won't want to miss this event. Visit the event website to view the agenda, download the event brochure
Learn More | Register to Attend
With the FREE Visitor Pass, you will have the opportunity to meet, learn, and network anywhere on the expo floor. Leading retailers, etailers, and grocers from across the globe will be there!
If you would like to attend the additional seminars and conference events not included in the expo, please see the conference website for details. Discounts for the full conference are available for groups of 3 or more - for more information please send an email to william.horgan@terrapinn.com or +1 212 379 6320.
]]>ABSTRACT:
A chance constrained optimization problem involves multiple uncertain constraints, i.e. constraints with stochastic parameters, that are required to be satisfied with probability at least a pre-specified threshold. In a distributionally robust chance constrained program (DRCCP), the chance constraint is required to hold for all probability distributions of the stochastic parameters from a given family of distributions, called an ambiguity set. In this work, we consider DRCCP involving linear uncertain constraints and an ambiguity set specified by convex moment inequalities. In general, a DRCCP is nonconvex and hence NP-hard to optimize. We develop deterministic reformulations of such problems and establish sufficient conditions under which these formulations are convex and tractable. We further approximate DRCCP by a system of single chance constraints, one for each uncertain constraint. The tractability of such approximation has been “an open question” since 2006. We provide sufficient conditions under which the approximation set is equivalent to the feasible region of DRCCP and can be reformulated as a convex program. Finally, we present a chance constrained optimal power flow model to illustrate the proposed methodology.
]]>Title: How to choose what you lift
Abstract: One way to generate valid cuts (that is, valid inequalities) for a mixed-integer set is to first relax any integrality constraints and create a cut for the continuous relaxation. However, as this approach ignores integrality information, cuts generated in this way may be weak. To remedy this, one may hope to somehow reintroduce integrality information in an attempt to strengthen the cut. This process of reimposing integrality on a variable is referred to as lifting the variable. In this talk we explore the idea of lifting from the viewpoint of cut-generating functions. We examine questions such as "Does lifting one variable produce a stronger cut than lifting another?" and "How much strength is gained from lifting a single variable?" This work was done in collaboration with Santanu Dey and Amitabh Basu.
Amitabh Basu (http://www.ams.jhu.edu/~abasu9/)
Title: Understanding Deep Neural Networks with Rectified Linear Units
Abstract: In this paper we investigate the family of functions representable by deep neural networks (DNN) with rectified linear units (ReLU). We give the first-ever polynomial time (in the size of data) algorithm to train a ReLU DNN with one hidden layer to global optimality. This follows from our complete characterization of the ReLU DNN function class whereby we show that a function is representable by a ReLU DNN if and only if it is a continuous piecewise linear function. The main tool used to prove this characterization is an elegant result from tropical geometry. Further, for the case, we show that a single hidden layer suffices to express all piecewise linear functions, and we give tight bounds for the size of such a ReLU DNN. We follow up with gap results showing that there is a smoothly parameterized family of "hard" functions that lead to an exponential blow-up in size, if the number of layers is decreased by a small amount. An example consequence of our gap theorem is that for every natural number , there exists a function representable by a ReLU DNN with depth and total size , such that any ReLU DNN with depth at most will require at least total nodes. Finally, we construct a family of functions for (also smoothly parameterized), whose number of affine pieces scales exponentially with the dimension at any fixed size and depth. To the best of our knowledge, such a construction with exponential dependence on has not been achieved by previous families of "hard" functions in the neural nets literature. This construction utilizes the theory of zonotopes from polyhedral theory.
]]>
Abstract:
Modern problems arising in many domains are driving a need for more capable, state-of-the-art optimization tools. A sharp focus on performance and accuracy has appeared, for example, in science and engineering applications. In particular, we have seen a growth in studies related to Polynomial Optimization: a field with beautiful and deep theory, offering flexibility for modeling and high impact in diverse areas. In this talk we will explore theoretical and practical LP-based techniques for polynomial optimization problems. Motivated by the AC-OPF problem in Power Systems, we will review how "tree-like" sparsity can be exploited as a tool for analysis of the fundamental complexity of the problem, by showing LP formulations that can efficiently approximate such sparse problems. In addition, we will show a computationally practical approach for constructing such approximations on-the-fly. Our methods rely on the maturity of current LP technology; we believe these contributions are important for the development of manageable approaches to general polynomial optimization problems.
Bio: Gonzalo Muñoz is a PhD Candidate of the Industrial Engineering and Operations Research Department at Columbia University. His research interests fit in the category of Non-Linear Mixed-Integer Optimization, including both theoretical perspectives and implementation of efficient algorithms to address this type of problems. Recently, he has worked on efficient LP approximations to sparse polynomial problems. The main applications of these methodologies are drawn from Power Grid operations and Mining scheduling problems.
]]>ABSTRACT:
We present a framework for nonparametric density estimation in situations where the sample is supplemented by information and assumptions about shape, support, continuity, slope, location of modes, density values, etc. These supplements are incorporated as constraints that in conjunction
with a maximum likelihood criterion lead to constrained infinite-dimensional optimization problems that we formulate for the first time over spaces of semicontinuous functions. These spaces, when equipped with the hypo-distance, offer a series of advantages including simple conditions for existence of estimators and their limits and the guaranteed convergence of modes of densities. Relying on epi-convergence, we provide general conditions under which estimators subject to nearly arbitrary constraints are consistent and illustrate the framework with a number of examples that span classical and novel shape constraints.
BIO:Dr. Johannes O. Royset is Associate Chair of Research and Professor of Operations Research at the Naval Postgraduate School. Prof. Royset's research focuses on formulating and solving stochastic optimization and variational problems arising in data science, sensor management, and engineering design. He was awarded a National Research Council postdoctoral fellowship in 2003, a Young Investigator Award from the Air Force Office of Scientific Research in 2007, and the Barchi Prize as well as the MOR Journal Award from the Military Operations Research Society in 2009. He received the Carl E. and Jessie W. Menneken Faculty Award for Excellence in Scientific Research in 2010 and was a co-recipient of the UPS George D. Smith Prize from INFORMS in 2013. He was a plenary speaker at the 14th International Conference on Stochastic Programming (2016). Prof. Royset is a Guest Editor of Mathematical Programming and an Associate Editor of Operations Research, Naval Research Logistics, Journal of Optimization Theory and Applications, and Computational Optimization and Applications. In 2015-2016, he was a Guest Editor of Journal of Optimization Theory and Applications. His research has been supported by the Office of Naval Research, Air Force Office of Scientific Research, Army Research Office, and DARPA and has resulted in one book, three book chapters, and 45 journal publications. He has a Doctor of Philosophy degree from the University of California at Berkeley (2002).
]]>
Abstract: Spatiotemporal data is ubiquitous in our daily life, including climate, transportation, and social media. Today, data is being collected at an unprecedented scale.
Yesterday’s concepts and tools are insufficient to serve tomorrow’s data-driven decision makers. Particularly, spatiotemporal data often demonstrates complex dependency structures and is of high dimensionality. This requires new machine learning algorithms that can handle highly correlated samples, perform efficient dimension reduction, and generate structured predictions.
In this talk, I will present tensor methods, a general framework for capturing high-order structures in spatiotemporal data. I will demonstrate how to learn from spatiotemporal data efficiently in both offline and online setting. I will also show interesting discoveries by our methods in climate and social media applications.
Bio: Qi (Rose) Yu is a Ph.D. candidate and Annenberg fellow at the University of Southern California focusing on machine learning and data analytics. Her research strives to develop machine learning methods to learn from large-scale spatiotemporal data, specifically in the domain of computational sustainability and social science. She has over a dozen publications in leading machine learning/data mining conferences and was nominated as one of the ``2015 MIT Rising Stars in EECS’’.
]]>
Abstract
Motivated by the applications in routing in data centers, we study the problem of expressing a doubly stochastic matrix as a linear combination using the smallest number of (sub)permutation matrices. The Birkhoff-von Neumann decomposition theorem proves the existence of such a decomposition, but does not give a representation with the smallest number of permutation matrices. In this talk, I will discuss the tractability of this problem from an exact and an approximate viewpoint. This is joint work with Janardhan Kulkarni and Euiwoong Lee.
]]>
Abstract: Heavy-tailed distributions are ubiquitous in modern statistical analysis and machine learning problems. This talk gives a simple principle for robust high-dimensional statistical inference via an appropriate shrinkage on the data. This widens the scope of high-dimensional techniques, reducing the moment conditions from sub-exponential or sub-Gaussian distributions to merely bounded second moment. As an illustration of this principle, we focus on robust estimation of the low-rank matrix from the trace regression model. It encompasses four popular problems: sparse linear models, compressed sensing, matrix completion, and multi-task regression. Under only bounded $2+\delta$ moment condition, the proposed robust methodology yields an estimator that possesses the same statistical error rates as previous literature with sub-Gaussian errors. We also illustrate the idea for estimation of large covariance matrix. The benefits of shrinkage are also demonstrated by financial, economic, and simulated data. Joint work with Weichen Wang and Zhiwei Zhu.
Bio: Jianqing Fan is Frederick L. Moore Professor at Princeton University. After receiving his Ph.D. from the University of California at Berkeley, he has been appointed as assistant, associate, and full professor at the University of North Carolina at Chapel Hill (1989-2003), professor at the University of California at Los Angeles (1997-2000), and professor at the Princeton University (2003--). He was the past president of the Institute of Mathematical Statistics and International Chinese Statistical Association. He is co-editing Journal of Econometrics and was the co-editor of The Annals of Statistics, Probability Theory and Related Fields and Econometrics Journal. His published work on statistics, economics, finance, and computational biology has been recognized by The 2000 COPSS Presidents' Award, The 2007 Morningside Gold Medal of Applied Mathematics, Guggenheim Fellow, P.L. Hsu Prize, Royal Statistical Society Guy medal in silver, and election to Academician of Academia Sinica and follow of American Associations for Advancement of Science.
]]>
Dietrich is an IBM Fellow and Vice President. She holds a BS in Mathematics from UNC and an MS and Ph.D. in OR/IE from Cornell. She joined IBM in 1984 and has worked in the area now called analytics for her entire career, applying data and computation to business decision processes throughout IBM. For over a decade, she led the Mathematical Sciences function in the IBM Research division where she was responsible for both basic research on computational mathematics and for the development of novel applications of mathematics for both IBM and its clients. She has been the president of INFORMS, has served on the Board of Trustees of SIAM, and is a member of several university advisory boards. She was elected to the National Academy of Engineering in 2014.
Her talk “Riding Technology Waves: Perspectives on the Deployment of Operations Research Methods” includes a fly-by of five decades of information technology beginning with its use to automate business processes and extending to its current role in intermediating social processes. The resulting "data exhaust," together with the availability of low cost computing capacity, spawned the age of analytics, the rise of big data, and the birth of cognitive computing. The past, current, and potential role of operations research in these technology waves will be discussed.
]]>
Abstract:
High dimensional estimation problems, such as phase retrieval, low rank matrix estimation, and blind deconvolution, attracted intensive attention recently due to their wide applications in medical image, signal processing, social networks, etc. Traditional approaches to solving these problems are either empirical, which work well but lack theoretic guarantee; or via convex formulations, which come with performance guarantee but are computationally costly in large dimensions. Nonconvex approaches are recently emerging as a powerful method to solve such problems, which come with provable performance guarantee and are computationally efficient.
In this talk, I will first introduce general ideas of using nonconvex methods for solving high dimensional estimation problems. I will then focus on the phase retrieval problem to present our recent advancements of nonconvex method. In particular, I will first describe our design of a nonconvex objective that yields first-order algorithm outperforming all existing algorithms in both statistical and computational efficiency. I will then present our further design of stochastic algorithms for large-scale phase retrieval with provable convergence guarantee. Towards the end of the talk, I will discuss insights learned from our studies, which are beneficial to future directions of this topic.
Bio: Dr. Yingbin Liang received the Ph.D. degree in Electrical Engineering from the University of Illinois at Urbana-Champaign in 2005. In 2005-2007, she was working as a postdoctoral research associate at Princeton University. In 2008-2009, she was an assistant professor at the University of Hawaii. Since December 2009, she has been on the faculty at Syracuse University, where she is an associate professor. Dr. Liang's research interests include information theory, statistical learning theory, optimization for large scale machine learning, and wireless communication and networks.
Dr. Liang was a Vodafone Fellow at the University of Illinois at Urbana-Champaign during 2003-2005, and received the Vodafone-U.S. Foundation Fellows Initiative Research Merit Award in 2005. She also received the M. E. Van Valkenburg Graduate Research Award from the ECE department, University of Illinois at Urbana-Champaign, in 2005. In 2009, she received the National Science Foundation CAREER Award, and the State of Hawaii Governor Innovation Award. In 2014, she received EURASIP Best Paper Award for the EURASIP Journal on Wireless Communications and Networking. She served as an Associate Editor for the Shannon Theory of the IEEE Transactions on Information Theory during 2013-2015.
]]>
Time: Monday, 29 May 2017 – 7:30am to Tuesday, 30 May 30 2017 – 5:00pm
Location: Georgia Tech Hotel
Event Details:
The Statistical Perspectives of Uncertainty Quantification workshop will focus on statistical and mathematical aspects of uncertainty quantification (UQ) of complex computational models.
Related Link:
]]>This course provides a deeper understanding of the PMBOK knowledge areas of project integration and procurement applied in the supply chain vendor selection and management process. To keep pace with the continuous moves toward outsourcing of operations and the advancement of technology, companies need to focus on selecting the right suppliers and partnerships to provide the most value to their customers and to remain profitable. This course provides the knowledge, skills, and tools to ensure that you are selecting the right supply chain partners (including 3PL’s) based on your business goals. Emphasis is placed on understanding alternative techniques for supplier selection including applied quantitative decision making techniques.
Complex supply chain transformation requires managing resources from many different departments, ensuring internal and external stakeholder alignment, mitigating large amounts of risk, and implementing communication, risk mitigation, and change management plans to ensure a successful project. Successful project management in complex supply chain environments requires application of well-planned integrated approaches. This course conveys an integrated view to supply chain transformation incorporating elements of change management, test plan development, project management techniques, and establishing effective project management teams.
Abstract:
In this talk, we will present a framework for analyzing, in the high-dimensional limit, the exact dynamics of a family of online algorithms for signal estimation and learning. For concreteness, we consider two prototypical problems: sparse principal component analysis and regularized linear regression (e.g. LASSO). For each case, we show that the time-varying estimates given by the algorithms will converge weakly to a deterministic “limiting process” in the high-dimensional limit. Moreover, this limiting process can be characterized as the unique solution of a nonlinear PDE, and it provides exact information regarding the asymptotic performance of the algorithms. For example, performance metrics such as the MSE, the cosine similarity and the misclassification rate in sparse support recovery can all be obtained by examining the deterministic limiting process. A steady-state analysis of the nonlinear PDE also reveals interesting phase transition phenomena related to the performance of the algorithms. Although our analysis is asymptotic in nature, numerical simulations show that the theoretical predictions are accurate for moderate signal dimensions.
Bio:
Yue M. Lu was born in Shanghai. After finishing undergraduate studies at Shanghai Jiao Tong University, he attended the University of Illinois at Urbana-Champaign, where he received the M.Sc. degree in mathematics and the Ph.D. degree in electrical engineering, both in 2007. Following his work as a postdoctoral researcher at the Audiovisual Communications Laboratory at Ecole Polytechnique Fédérale de Lausanne (EPFL), Switzerland, he joined Harvard University in 2010, where he is currently an Associate Professor of Electrical Engineering at the John A. Paulson School of Engineering and Applied Sciences.
He received the Most Innovative Paper Award of IEEE International Conference on Image Processing (ICIP) in 2006, and the Best Student Paper Award of IEEE ICIP in 2007. Student papers supervised and coauthored by him won the Best Student Paper Award (with Ivan Dokmanic and Martin Vetterli) of IEEE International Conference on Acoustics, Speech and Signal Processing in 2011 and the Best Student Paper Award (with Ameya Agaskar and Chuang Wang) of IEEE Global Conference on Signal and Information Processing (GlobalSIP) in 2014. He received the ECE Illinois Young Alumni Achievement Award in 2015.
]]>
Plan for Every Part (PFEP) is proven to be the most effective way to optimize end to end material flow. Breakthrough results in working capital improvement, inventory right-sizing, on-time-delivery, manufacturing throughput, and cost reduction are achieved through this solution, but how do you do it?
This series will work backwards from customer demand to manufacturing conveyance and presentation to supplier replenishment methods and frequencies in order to develop, deploy and sustain end to end lean material flow processes. You will learn the critical components of PFEP that ensure delivery of the right parts, at the right time, in the right place, in the right quantity, with the right quality, at the right cost.
Course 1: Introduction to PFEP and Inventory Sizing
How can my organization use PFEP to optimize material flow? What does an effective PFEP program look like? How should we size inventory and present material to production?
Course 2: Plan for Every Part (PFEP) and Inventory Layout is May 18-19, 2017 (you may want to stay two additional days and register for that course to complete both in one week)
Executive through manager-level leaders responsible for working capital improvement, supply chain, logistics, materials, purchasing, production, warehousing, and transportation.
The course will cover the below content:
Participants receive a course notebook and hand-out materials.
None. This course is the first of the three-course Plan for Every Part (PFEP) series.
This series works backwards from customer demand to manufacturing conveyance and presentation to supplier replenishment methods and frequencies in order to develop, deploy and sustain end to end lean material flow processes. You will learn the critical components of PFEP that ensure delivery of the right parts, at the right time, in the right place, in the right quantity, with the right quality, at the right cost.
Standard: $1,450.00, Alumni/Org Discount: $1,305.00, Certificate: $1,203.50 (cost of each course when signing up for and paying for a multi-course certificate program).
First time attendees pay the listed course fee. If you are a returning student of the Supply Chain & Logistics Institute (SCL) courses, you will receive a 10% discount which you will enter at the "Check Out" page. Use Coupon Code SCL-Alum.
There are also discounts available for multiple-team member registrations, to those who prepay for all the courses in a specific certificate, to active/retired military, or to members of certain organizations (click this link for a listing).
Discounts cannot be combined. To receive the coupon code for these discounts, call 404-385-8663 or send us an email prior to registration
The program fee for LIVE courses (non-online) includes continental breakfasts, lunches, breaks, parking, internet access, and all classroom materials.
If the Supply Chain & Logistics Institute must cancel a program, registrants will receive a full refund. Georgia Tech, however, cannot assume the responsibility for other costs incurred. Due to program enrollment limits, early registration is encouraged. Registrations will be acknowledged by a letter of confirmation from Professional Education.
On the first day, please check in at least 30 minutes before the class start time.
Every effort is made to present the course as advertised herein; however, circumstances may make it necessary to alter the schedule and/or presenters.
This series will work backwards from customer demand to manufacturing conveyance and presentation to supplier replenishment methods and frequencies in order to develop, deploy and sustain end to end lean material flow processes. You will learn the critical components of PFEP that ensure delivery of the right parts, at the right time, in the right place, in the right quantity, with the right quality, at the right cost.
]]>Plan for Every Part (PFEP) is proven to be the most effective way to optimize end to end material flow. Breakthrough results in working capital improvement, inventory right-sizing, on-time-delivery, manufacturing throughput, and cost reduction are achieved through this solution, but how do you do it?
This series will work backwards from customer demand to manufacturing conveyance and presentation to supplier replenishment methods and frequencies in order to develop, deploy and sustain end to end lean material flow processes. You will learn the critical components of PFEP that ensure delivery of the right parts, at the right time, in the right place, in the right quantity, with the right quality, at the right cost.
Course 3: PFEP and Total Cost Management
How can my organization optimize inbound logistics to support a PFEP program? How can it source suppliers and optimize material flow? How can our leaders measure and sustain a PFEP program? How do we manage the total cost of our supply chain?
Executive through manager-level leaders responsible for working capital improvement, supply chain, logistics, materials, purchasing, production, warehousing, and transportation.
The course will cover the below content:
Participants receive a course notebook and hand-out materials.
None. This course is the third and last of the three-course Plan for Every Part (PFEP) series.
This series works backwards from customer demand to manufacturing conveyance and presentation to supplier replenishment methods and frequencies in order to develop, deploy and sustain end to end lean material flow processes. You will learn the critical components of PFEP that ensure delivery of the right parts, at the right time, in the right place, in the right quantity, with the right quality, at the right cost.
Standard: $1,450.00, Alumni/Org Discount: $1,305.00, Certificate: $1,203.50 (cost of each course when signing up for and paying for a multi-course certificate program).
First time attendees pay the listed course fee. If you are a returning student of the Supply Chain & Logistics Institute (SCL) courses, you will receive a 10% discount which you will enter at the "Check Out" page. Use Coupon Code SCL-Alum.
There are also discounts available for multiple-team member registrations, to those who prepay for all the courses in a specific certificate, to active/retired military, or to members of certain organizations (click this link for a listing).
Discounts cannot be combined. To receive the coupon code for these discounts, call 404-385-8663 or send us an email prior to registration
The program fee for LIVE courses (non-online) includes continental breakfasts, lunches, breaks, parking, internet access, and all classroom materials.
If the Supply Chain & Logistics Institute must cancel a program, registrants will receive a full refund. Georgia Tech, however, cannot assume the responsibility for other costs incurred. Due to program enrollment limits, early registration is encouraged. Registrations will be acknowledged by a letter of confirmation from Professional Education.
On the first day, please check in at least 30 minutes before the class start time.
Every effort is made to present the course as advertised herein; however, circumstances may make it necessary to alter the schedule and/or presenters.
This series will work backwards from customer demand to manufacturing conveyance and presentation to supplier replenishment methods and frequencies in order to develop, deploy and sustain end to end lean material flow processes. You will learn the critical components of PFEP that ensure delivery of the right parts, at the right time, in the right place, in the right quantity, with the right quality, at the right cost.
]]>
ABSTRACT:
Optimization problems arising in decentralized multi-agent systems have gained significant attention in the context of cyber-physical, communication, power, and robotic networks combined with privacy preservation, distributed data mining and processing issues. The distributed nature of the problems is inherent due to partial knowledge of the problem data (i.e., a portion of the cost function or a subset of the constraints is known to different entities in the system), which necessitates costly communications among neighboring agents. In this talk, we present a new class of decentralized first-order methods for nonsmooth and stochastic optimization problems which can significantly reduce the number of inter-node communications. Our major contribution is the development of decentralized communication sliding methods, which can skip inter-node communications while agents solve the primal subproblems iteratively through linearizations of their local objective functions.
Bio
Soomin Lee is a postdoc fellow in industrial and systems engineering at Georgia Tech. She received her Ph.D. in Electrical and Computer Engineering (2013) and Master's in Computer Science (2012) from the University of Illinois, Urbana-Champaign. After graduation, she joined in Duke Robotics Group as a postdoc associate. In 2009, she worked as an assistant research officer at the Advanced Digital Science Center (ADSC) in Singapore. She is a recipient of the NSF fellowship program for enhancing partnership with industry. Her research interests include control and optimization of various distributed engineering systems interconnected over complex networks, large-scale machine learning for big data analytics as well as theoretical optimization.
]]>
ABSTRACT:
Hydropower producers rely on stochastic optimization in their daily scheduling of resources. Due to uncertainties in future inflow to reservoirs and market prices, stochastic optimization models often proves to outperform their deterministic and heuristic counterparts. In this presentation the combined SDP/SDDP methodology widely used by Nordic producers will be described. We discuss how both changes in power market design as well as the future generation mix in Europe introduces new challenges to this methodology. In particular it will be of higher importance for the methodology to capture the exact unit commitment of generators and to incorporate the uncertainty in multiple price processes.
]]>
***Supply Chain Executive Forum***
SCL Corporate Partners, Advisory Board Members, and Faculty are invited to attend and public seats will also be available. The session will take place 4:30-6pm on Day 1 of the conference. Our topic for the session will be "Network Design Challenges in an eCommerce/Omni-Channel Environment". Speakers from leading retailers, manufacturers, and academia will discuss their challenges and approaches to dealing with the dynamic changes associated with growing e-commerce and omni-channel distribution networks. The audience will be engaged in discussion regarding the challenges and emerging solutions to address the complexities of multi-point, multi-channel forecasting, inventory deployment, and facility and transportation planning. Opportunities for collaborative research efforts will also be discussed.
***Workshop on the Role of Labels in the Growing eCommerce World***
1-2pm on Day 1 of the conference.
***Logistics Technology Disruptors***
Edenfield Executive in Residence, Moe Trebuchon, will be speaking as part of the session from 11:15-12:15pm on Day 1 of the conference.
We hope to see you at our booth or our sessions. SCL partners, please email event@scl.gatech.edu for registration details.
]]>Blending personal and professional life experiences, Robert addresses current-day challenges as well as opportunities that exist in our new world of constant disruption.
In the end, Robert shares wisdom and knowledge that will allow today’s business leaders to continue down the path to successfully discover hidden profit.
ABOUT OUR SPEAKER
Robert Martichenko is the founder and CEO of LeanCor Supply Chain Group, a trusted supply chain partner with a mission to advance the world’s supply chains through training, consulting, and third party logistics.
Formally educated in mathematics, history and finance, Robert is a globally recognized thought leader in lean thinking and end-to-end supply chain management; an award-winning business author, successful novelist and recipient of the 2015 Distinguished Service Award from the Council of Supply Chain Management Professionals.
In addition, and most importantly, Robert is a father, husband and friend.
Robert’s interests, expertise, and speaking topics bridge the often separate worlds of personal, professional and organizational growth. Robert is an engaging, trending and thought-provoking speaker; allowing audiences to walk away feeling energized, enlightened and optimistic about opportunities that lay ahead.
Please note that registration is required via the below link. Lunch will be provided to registered attendees.
]]>
ABSTRACT:
In recent years, the paradigm of cloud computing has emerged as an architecture for computing that makes use of distributed (networked) computing resources. In this paper, we consider a distributed computing algorithmic scheme for stochastic optimization which relies on modest communication requirements amongst processors and most importantly, does not require synchronization. Specifically, we analyze a scheme with N > 1 independent threads implementing each a stochastic gradient algorithm. The threads are coupled via a perturbation of the gradient (with attractive and repulsive forces) in a similar manner to mathematical models of flocking, swarming and other group formations found in nature with mild communication requirements. When the objective function is convex, we show that a flocking-like approach for distributed stochastic optimization provides a noise reduction effect similar to that of a single-thread stochastic gradient algorithm based upon the average of N gradient samples at each step. The distributed nature of flocking makes it an appealing computational alternative. We show that when the overhead related to the time needed to gather N samples and synchronization is not negligible, the flocking implementation outperforms a single-thread stochastic gradient algorithm based upon the average of N gradient samples at each step. When the objective function is not convex, the flocking-based approach seems better suited to escape locally optimal solutions due to the repulsive force which enforces a certain level of diversity in the set of candidate solutions. Here again, we show that the noise reduction effect is similar to that associated to the single-thread stochastic gradient algorithm based upon the average of N gradient samples at each step.
Bio:
Alfredo Garcia is Professor with the Department of Industrial and Systems Engineering at the University of Florida. He received an undergraduate degree in Electrical Engineering from the Universidad de los Andes, Colombia, in 1991, the Diplome d'Etudes Approfondies D.E.A. in Control Systems from the Université Paul Sabatier, Toulouse, France, in 1992, and the Ph.D. degree in Operations Research from the University of Michigan, Ann Arbor, in 1997. During 1998-2000 he served as Commissioner in the Colombian Energy Regulatory Commission. From 2001-2015 he was a member of the faculty at the University of Virginia. His research interests include game theory and dynamic optimization with applications in power and communication networks. He currently manages the program in "Control of Networked Multi-agent Systems" for ARO.
]]>
ABSTRACT:
We study cake cutting on a graph, where agents can only evaluate their shares relative to their neighbors. This is an extension of the classical problem of fair division to incorporate the notion of social comparison from the social sciences. We say an allocation is locally envy-free if no agent envies a neighbor’s allocation, and locally proportional if each agent values its own allocation as much as the average value of its neighbors’ allocations. We generalize the classical “Cut and Choose” protocol for two agents to this setting, by fully characterizing the set of graphs for which an oblivious single-cutter protocol can give locally envy-free (thus also locally-proportional) allocations. We study the price of envy-freeness, which compares the total value of an optimal allocation with that of an optimal, locally envy-free allocation. Surprisingly, a lower bound of Ω(√n) on the price of envy-freeness for global allocations also holds for local envy-freeness in any connected graph, so sparse graphs do not provide more flexibility asymptotically with respect to the quality of envy-free allocations.
Bio:
Rediet Abebe is a PhD student in the Department of Computer Science at Cornell University, advised by Jon Kleinberg. Her research focuses on algorithms, computational social science, and social networks. In particular, she is interested in using insights from theoretical computer science to better understand and implement interventions in socioeconomic inequality and opinion dynamics. She is the 2016 recipient of the Google Generation Scholarship. Prior to Cornell, she completed a B.A. in Mathematics from Harvard University, an M.A. in Mathematics from the University of Cambridge, and an M.S. in Applied Mathematics from Harvard University. She was born and raised in Addis Ababa, Ethiopia.
]]>
ABSTRACT:
Modern manufacturing needs to optimize the entire product lifecycle to satisfy the highly diverse customer needs. With the deployment of Industrial Internet and sensor/actuator networks, data-driven decision making is expected to enable smart manufacturing to achieve high level of adaptability and flexibility. Such a manufacturing system generates spatially and temporally dense data sets. This talk focuses on data-driven modeling problems with functional data, where the models will be used in data-driven decision making in smart manufacturing. Examples in functional variable selections, in situ process modeling, and data interpretation from natural language processing perspective will be discussed in this talk. The methodology has been broadly applied to many advanced manufacturing processes, such as aero-engine manufacturing, crystal growth manufacturing, and additive manufacturing.
Bio. Dr. Ran Jin is an assistant professor and the Director of Laboratory of Data Science and Visualization at the Grado Department of Industrial and Systems Engineering at Virginia Tech. He received his Ph.D degree in Industrial Engineering from Georgia Tech. He worked with Prof. Jianjun(Jan) Shi on multistage manufacturing research in the System Informatics and Control group. After his graduation, his research focuses on Data Fusion in Smart Manufacturing, including the integration of different types of data sets (e.g., ensemble models), variables (e.g., quantitative and qualitative models), and information (e.g., product quality and equipment reliability) for synergistically modeling, monitoring and control of manufacturing processes and systems. For more information about Dr. Jin, please visit: http://www.ise.vt.edu/People/Faculty/Bios/JinRan_bio.html
]]>
Abstract:
We live in an increasingly data-driven world in which mathematical models are crucial for uncovering properties of systems from measured data. Graphical models are commonly used for capturing the relationships between the parameters of a system using graphs. Graphical models have applications in many areas, such as social sciences, linguistics, neuroscience, biology, and power systems. Learning graphical models is of fundamental importance in machine learning and statistics, and is often challenged by the fact that only a small number of samples are available. Several algorithms (such as Graphical Lasso) have been proposed to address this problem. Despite the popularity of graphical lasso, there is not much known about the properties of this statistical method as an optimization algorithm. In this talk, we will develop new notions of sign-consistent matrices and inverse consistent matrices to obtain key properties of graphical lasso. In particular, we will prove that although the complexity of solving graphical lasso is high, the sparsity pattern of its solution has a simple formula if a sparse graphical model is sought. Besides graphical lasso, there are several other techniques for learning graphical models. However, it is not clear how reliable these methods are and which method should be used for each particular application. To address these problems, we will design a novel framework for generating synthetic data based on stochastic electrical circuits, and use it as a platform to assess the performance of various techniques. We will show that our platform can be used to first find the best algorithm and then identify the best model by optimally adjusting the controllable parameters of the algorithm. We will illustrate our results on fMRI data and uncover new properties of brain networks.
Bio: Somayeh Sojoudi is an Assistant Project Scientist at the University of California, Berkeley. She received her PhD degree in Control & Dynamical Systems from California Institute of Technology in 2013. She was an Assistant Research Scientist at New York University School of Medicine from 2013 to 2015. She has worked on several interdisciplinary problems in optimization, control theory, machine learning, data analytics, and power systems. Somayeh Sojoudi is an associate editor for the IEEE Transactions on Smart Grid. She is a co-recipient of the 2015 INFORMS Optimization Society Prize for Young Researchers and a co-recipient of the 2016 INFORMS ENRE Energy Best Publication Award. She is a co-author of a best student paper award finalist for the 53rd IEEE Conference on Decision and Control 2014.
]]>
This course provides a deeper understanding of the PMBOK knowledge areas of project integration and procurement applied in the supply chain vendor selection and management process. To keep pace with the continuous moves toward outsourcing of operations and the advancement of technology, companies need to focus on selecting the right suppliers and partnerships to provide the most value to their customers and to remain profitable. This course provides the knowledge, skills, and tools to ensure that you are selecting the right supply chain partners (including 3PL’s) based on your business goals. Emphasis is placed on understanding alternative techniques for supplier selection including applied quantitative decision making techniques.
Complex supply chain transformation requires managing resources from many different departments, ensuring internal and external stakeholder alignment, mitigating large amounts of risk, and implementing communication, risk mitigation, and change management plans to ensure a successful project. Successful project management in complex supply chain environments requires application of well-planned integrated approaches. This course conveys an integrated view to supply chain transformation incorporating elements of change management, test plan development, project management techniques, and establishing effective project management teams.
ABSTRACT:
Simulation is a powerful numerical tool set for performance evaluation and optimization of stochastic systems. Successful implementation of this numerical approximation scheme requires one being able to assess the quality of the estimators and control the estimation errors. In this talk, I will present a new sequential stopping framework for stochastic simulation problems in which variance estimation is difficult. Examples include estimating smooth function of expectations, steady-state simulation, and various stochastic optimization problems. The proposed procedure guarantees the accuracy of the estimator and achieves the desired reliability asymptotically. (Joint work with Peter Glynn.)
Bio: Jing Dong is an assistant Professor in the Department of Industrial Engineering and Management Sciences at Northwestern University. Her research interests are applied probability, stochastic simulation, and service operations management. She got her Phd in Operations Research from Columbia University.
]]>Supply Chain Management projects can span a wide range of project types including supply chain strategy, network analysis, facility design build, supply chain technology selection and implementation, and continuous process improvement initiatives. This course provides an overview of project management methodologies as applied in the supply chain environment. Class discussion and projects provide an understanding of how fundamental project management approaches and industry best practices can be used to effectively manage the complexities. Supply chain projects typically require managing resources, stakeholder alignment, risk management, customer impact, and effective communication across many internal and external business partners.
Upon completion of this course, you will be able to:
None.
For those interested in earning the Supply Chain Project Management Certificate, this course is the first of the four-course certificate program. To earn the certificate, participants must register and complete the following courses in the below sequence within four years, plus one elective.
For a list of courses that can be used as electives towards this certificate, please visit the Georgia Tech Professional Education website.
]]>Supply Chain Management projects can span a wide range of project types including supply chain strategy, network analysis, facility design build, supply chain technology selection and implementation, and continuous process improvement initiatives. This course provides an overview of project management methodologies as applied in the supply chain environment. Class discussion and projects provide an understanding of how fundamental project management approaches and industry best practices can be used to effectively manage the complexities. Supply chain projects typically require managing resources, stakeholder alignment, risk management, customer impact, and effective communication across many internal and external business partners.
Upon completion of this course, you will be able to:
None.
For those interested in earning the Supply Chain Project Management Certificate, this course is the first of the four-course certificate program. To earn the certificate, participants must register and complete the following courses in the below sequence within four years, plus one elective.
For a list of courses that can be used as electives towards this certificate, please visit the Georgia Tech Professional Education website.
]]>Do you work with problems involving the use of material handling equipment in plants, warehouses and other commercial enterprises? Focus on material handling and distribution problems from the source of raw material through manufacturing and distribution systems to the final consumer. All techniques presented are field-proven and derived from successful implementation. Case exercises are adapted from real situations and projects.
Industrial engineers and systems analysts, warehouse supervisors and team leaders, warehouse/distribution center managers, logistics and supply chain planners, planning teams for new or expanded facilities and leaders of supply chain and lean initiatives
Upon completion of this course, you will be able to:
Participants receive a course notebook.
COURSE PREREQUISITES
None.
This course is part of the Distribution Operations Analysis and Design (DOAD) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
ABSTRACT:
Predictions about the future are a crucial part of the decision making process in many real-world online problems. However, analysis of online algorithms has little to say about how to use predictions, and how properties of prediction errors impact algorithm design. In this talk, I'll describe recent results exploring the power of predictions in online convex optimization and how properties of prediction noise can impact the structure of optimal online algorithms. I will also briefly highlight applications of these tools to data centers and the smart grid.
Bio:
Adam Wierman is a Professor in the Department of Computing and Mathematical Sciences at the California Institute of Technology, where he currently serves as Executive Officer. He is also the director of the Information Science and Technology (IST) initiative at Caltech. He is the founding director of the Rigorous Systems Research Group (RSRG) and co-Director of the Social and Information Sciences Laboratory (SISL). His research interests center around resource allocation and scheduling decisions in computer systems and services. He received the 2011 ACM SIGMETRICS Rising Star award, the 2014 IEEE Communications Society William R. Bennett Prize, and has been coauthor on papers that received of best paper awards at ACM SIGMETRICS, IEEE INFOCOM, IFIP Performance (twice), IEEE Green Computing Conference, IEEE Power & Energy Society General Meeting, and ACM GREENMETRICS. Additionally, he maintains a popular blog called Rigor + Relevance.
]]>ABSTRACT:
Distributed systems are ubiquitous in both industry and our daily life. For example, we use clusters and networked workstations to analyze large amount of data, use the world wide web for information and resource sharing, and use the Internet of Things (IoT) to access a much wider variety of resources. In distributed systems, components are more vulnerable to adversarial attacks.
In this talk, I model the distributed systems as multi-agent networks, and consider the most general attack model – Byzantine fault model. In particu- lar, I will focus on the problem of distributed learning over multi-agent net- works, where agents repeatedly collect partially informative observations (sam- ples) about an unknown state of the world, and try to collaboratively learn the true state. We focus on the impact of the Byzantine agents on the performance of consensus-based non-Bayesian learning. Our goal is to design algorithms for the non-faulty agents to collaboratively learn the true state through local communication.
At the end of this talk, I will also briefly mention my exploration on tolerating adversarial attacks in multi-agent optimization problems.
Lili Su is a Ph.D. candidate in the Electrical and Computer Engineering Department at the University of Illinois at Urbana-Champaign, working with Prof. Nitin Vaidya on distributed computing. She expects to receive her Ph.D. degree in May 2017.
Her research intersects distributed computing, security, optimization, and learning. She was one of the three nominees for the 2016 International Sym- posium on DIStributed Computing Best Student Paper Award. She received the 2015 International Symposium on Stabilization, Safety, and Security of Dis- tributed Systems Best Student Paper Award. She also received the Sundaram Seshu International Student Fellowship for the academic year of 2016 to 2017 conferred by UIUC. In addition, she received the Outstanding Reviewer Award for her review service for IEEE Transactions on Communication in 2015.
]]>ABSTRACT:
Steady-state diffusion approximations are commonly used to approximate models of large scale service systems. In this talk I will introduce a framework based on Stein's method that
can be used a) as an engineering solution for generating good steady-state approximations
and b) as a mathematical tool for establishing error bounds for these approximations. These approximations are often universally accurate in multiple parameter regions, from underloaded, to critically loaded, to overloaded (when customers abandon). As a running example, I will use the many server queue with customer abandonment and phase-type service time distribution,
which is a fundamental building block in service system models.
Bio: Anton is currently a Ph.D student at Cornell University in the Operations Research Department working with Professor Jim Dai. He received a Bachelors degree in Math and Statistics from the University of Toronto. Broadly speaking, He is interested in stochastic modeling and applied probability. His thesis work focuses on applying Stein's method to the study of steady-state approximations of stochastic systems. These approximations include both first order approximations such as mean-field/fluid approximations, and second order approximations such as diffusion approximations. However, he is also interested in Markov decision processes and stochastic control theory. With respect to application domains, he is interested in fleet management questions in ridesharing systems such as Uber, Lyft, or Didi.
]]>ABSTRACT:
Input-output uncertainty (IOU) comparisons is a new discrete optimization via simulation (DOvS) procedure for a problem under input model risk. Input model risk refers to the risk of making a suboptimal decision due to misspecification of input models to the simulation because they are estimated from finite real-world observations. Without knowing the true input distributions, finding the true optimal solution is a challenging, yet ubiquitous problem. IOU comparisons procedure separates the solutions in contention into two groups with the target probability guarantee; the solutions so inferior to the true optimal that we can separate them even with the errors in the input models and the solutions statistically inseparable from the true optimal with the input model errors. The size of the latter subset is small, if not one, when the solutions are affected similarly by the input model errors and/or when the true optimal solution performs much better than the rest. IOU comparisons procedure asymptotically provides the target statistical guarantee as the real-world sample size and simulation effort increase. The finite sample performances of the IOU comparisons procedure and its heuristic variation are demonstrated using an inventory problem with unknown input distributions.
Bio: Eunhye Song is a Ph.D. candidate in Industrial Engineering Management Sciences at Northwestern University. Her primary research interests lie in optimization via simulation and uncertainty quantification in simulation output due to misspecified input models. Her most recent research combines both areas to determine solutions which perform well even in the presence of model risk due to such misspecification.
A chapter of her dissertation is now implemented in the output analysis module of Simio, a leading simulation software package. She also has worked on an industry project with General Motors to quantify uncertainty in their market share simulator.
In her free time, she enjoys travelling to a new place, trying a new recipe she found on the internet, and playing squash with friends.
]]>ABSTRACT:
At the heart of most algorithms today, there is an optimization engine trying to provide the best decision with partial information observed thus far in time, the so-called problem of online learning. Often it becomes difficult to find near-optimal solutions to these problems due to their inherent combinatorial structure that leads to certain computational bottlenecks, for instance, computing Bregman projections for online mirror descent and its variants. Motivated by these bottlenecks, we consider three fundamental convex and combinatorial optimization problems. First, we provide a conceptually simple algorithm to minimize separable convex functions over submodular base polytopes. For cardinality-based submodular functions, we show the current fastest-known running time of O(n(log n+d)), where n is the size of the ground set and d is the number of distinct values of the submodular function (d<=n). Next, we consider the problem of movement along a line while staying feasible in submodular polytopes. The use of the discrete Newton’s method for this line search problem is natural, but no strongly polynomial bound on its number of iterations was known. We solve this open problem by providing a quadratic bound of O(n2), resulting in a running time improvement by at least n6 over the state of the art. Finally, we show how efficient counting methods can be used for convex minimization. This is joint work with Michel Goemans and Patrick Jaillet.
Bio: Swati Gupta is a PhD candidate in the Operations Research Center (ORC) and Laboratory for Information and Decision Systems (LIDS) at MIT, jointly advised by Michel Goemans and Patrick Jaillet. She graduated from Indian Institute of Technology (IIT), Delhi, in 2011 with dual Bachelors and Masters degree in Computer Science and Engineering. She won the Google Women in Engineering Award in 2011, received a special recognition from the INFORMS Computing Society in 2016 and was a finalist in the INFORMS Service Science student paper award in 2016.
]]>ABSTRACT:
Privacy concerns are becoming a major obstacle to using data in the way that we want. It's often unclear how current regulations should translate into technology, and the changing legal landscape surrounding privacy can cause valuable data to go unused. In addition, when people know that their current choices may have future consequences, they might modify their behavior to ensure that their data reveal less --- or perhaps, more favorable --- information about themselves. Given these concerns, how can we continue to make use of potentially sensitive data, while providing satisfactory privacy guarantees to the people whose data we are using? Answering this question requires an understanding of how people reason about their privacy and how privacy concerns affect behavior.
In this talk, we will see how strategic and human aspects of privacy interact with existing tools for data collection and analysis. I will begin by adapting the standard model of consumer choice theory to a setting where consumers are aware of, and have preferences over, the information revealed by their choices. In this model of privacy-aware choice, I will show that little can be inferred about a consumer's preferences once we introduce the possibility that she has concerns about privacy, even when her preferences are assumed to satisfy relatively strong structural properties. Next, I will analyze how privacy technologies affect behavior in a simple economic model of data-driven decision making. Intuition suggests that strengthening privacy protections will both increase utility for the individuals providing data and decrease usefulness of the computation. I will demonstrate that this intuition can fail when strategic concerns affect consumer behavior. Finally, I'll discuss ongoing behavioral experiments, designed to empirically measure how people trade off privacy for money, and to test whether human behavior is consistent with theoretical models for the value of privacy.
Bio: Rachel Cummings is a Ph.D. candidate in Computing and Mathematical Sciences at the California Institute of Technology. Her research interests lie primarily in data privacy, with connections to optimization, economics, decision-making, and information systems. Her work has focused on problems such as strategic aspects of data generation, incentivizing truthful reporting of data, privacy-preserving algorithm design, impacts of privacy policy, and human decision-making. She received her B.A. in Mathematics and Economics from the University of Southern California and her M.S. in Computer Science from Northwestern University. She won the Best Paper Award at the 2014 International Symposium on Distributed Computing, and she is the recipient of a Simons Award for Graduate Students in Theoretical Computer Science.
]]>In today’s global economy, operating risks are increasingly on the minds of executives. The specific context of operating risk can range from general areas of business continuity to the effects of natural disasters. In this course participants will gain a solid understanding of Supply Chain Risk Management principals including effective ways to identify, mitigate and measure the impact of potential supply chain disruptions.
Upon completion of this course, you will be able to:
Participants receive a course notebook, an in-class software demonstration, and hand-out materials.
No prerequisites. For those interested in earning the Supply and Demand Planning Certificate, take the below core courses and one elective course within four years. Please note that we are offering all three core courses in the series during the week of June 6, 2016 on the Georgia Tech campus.
Laptop computer by the student.
This course is part of the Supply and Demand Planning (SDP) Certificate.
PROGRAM TIMES
On the first day, please check in at least 30 minutes before the class start time.
Develop effective supply chain performance measurement processes to drive alignment across the corporation. Corporations often struggle with a lack of alignment between financial goals and operational metrics. Additionally, today’s information technology often overwhelms management with data and metrics. In this 3-day course you will learn how to develop metrics that synchronize supply chain and logistics metrics with key company financial metrics and goals. Through coursework and hands-on exercises, students learn to tailor metrics and measurement processes to focus on the most important aspects of the operations and overall corporate goals. As part of the course, performance dashboards from participating student companies and best-in-class organizations will be critiqued. The methods learned in the course are intended to be immediately put to use in the corporate environment.
Chief supply chain officers, executive vice presidents of supply chain, executive vice presidents of procurement, company owners, logistics service providers, consultants, vice presidents of sales operations, vice presidents/directors of process improvement, executive/senior/vice presidents/directors of supply chain, executive/senior/vice presidents/directors of logistics, executive/senior/vice presidents/directors of procurement, executive/senior/vice presidents/directors of manufacturing, executive/senior/vice presidents/directors of distribution
Upon completion of this course, you will be able to:
WHAT IS COVERED
Participants receive a course notebook.
COURSE PREREQUISITES
None.
This course is part of the Supply Chain Management (SCM) Certificate.
PROGRAM TIMES
On the first day, please check in at least 30 minutes before the class start time.
This course will demonstrate how warehouse operations are a key enabler to a successful supply chain implementation and the starting point for a transformation. It is critical to manage safety, quality and efficiency. Learn to leverage the lean supply chain modifications to improve customer responsiveness and reduce operating costs and in doing so contributing to a supply chain that creates a competitive advantage for a company. To accomplish this goal, we must bring lean principles into the warehouse and distribution center.
Supply chain professionals, logistics professionals, material managers, production control managers, transportation managers, warehousing managers and purchasing managers
Upon completion of this course, you will be able to:
Benefits:
Required
Provided
COURSE PREREQUISITES
None.
This course is part of the Distribution Operations Analysis and Design (DOAD) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
As supply chain executives become more instrumental in supporting long-term strategic objectives, they need to complement traditional supply chain operational knowledge with a more strategic view of their role in delivering aligned results to the business. Learn about different lenses of strategic planning applied to supply chain management and the specific implications on supply chain operations.
During the course, attendees will have several opportunities to assess their current supply chain strategy, formulate a new one, discuss about keys to implement a demand driven supply chain strategy, and how to review and align one.
For the duration of the course, participants will have the opportunity to work with an extended simulation game of a fictional company, and see the impact of supply chain strategic decisions in real time, while monitoring their ability to manage uncertainty and deliver financial results for the firm.
Those responsible for determining the future position of supply chain strategy, executing specific supply chain processes that must support specific business initiatives, innovating supply chain strategy to better align with operational goals, understanding the strategic impact of supply chain decisions in the firm and the extended enterprise network and advising clients about their specific supply chain strategic positioning.
Participants receive a course notebook.
None.
Laptop computer and calculator to be provided by the student.
This course is part of the Supply Chain Management (SCM) Certificate and can also be used towards the Strategic Sourcing and Supply Management (SSSM) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
This course provides a holistic view toward corporate profitability and supports effective complexity management. Participants will learn about the challenges of today's operating environment with "big data," cross-functional consensus and strategies that impact profitability. Integrated Business Planning (IBP) building blocks will be provided that solve these challenges. Break-out sessions will allow participants to apply these IBP concepts with an interactive tool.
Upon completion of this course, you will be able to:
Participants receive a course notebook and reading materials.
None.
Laptop computer by the student.
This course is part of the Supply and Demand Planning (SDP) Certificate.
PROGRAM TIMES
On the first day, please check in at least 30 minutes before the class start time.
This course focuses on defining, executing, and improving the S&OP process. Participants will be introduced to the appropriate stakeholders of S&OP, the importance of S&OP to corporate performance, S&OP cadence, and the use of visionary technology to bring S&OP to the next level. Business cases will be used to show concrete examples of companies where S&OP is effectively applied.
Upon completion of this course, you will be able to:
Participants receive a course notebook, an in-class software demonstration, and hand-out materials.
None.
Laptop computer by the student.
This course is part of the Supply and Demand Planning (SDP) Certificate.
PROGRAM TIMES
On the first day, please check in at least 30 minutes before the class start time.
Effective planning of transportation and distribution networks has become more complex. This is driven by increasing customer requirements, expansion of global sourcing, security and regulatory requirements, volatile fuel costs, etc. This course is focused on understanding the strategic and tactical principles, practices, and tools required to address the cost, service, capacity, and carbon emissions tradeoffs in domestic and international transportation.
Executives and managers who want to learn about designing and operating best-in-class transportation and distribution for their supply chain, supply chain engineers and analysts, industrial engineers and systems analysts, supply chain and logistics consultants, transportation managers and engineers, and operations managers of traffic, transportation or warehousing
Upon completion of this course, you will be able to:
Participants receive a course notebook.
None.
This course is part of the Supply Chain Management (SCM) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
Supply chain management and logistics encompasses all the activities involved in getting products to consumers including planning, storing, moving, and accounting for inventory. Inventory availability is the most important aspect of customer service, and the cost of inventory is one of the most important entries on a company's balance sheet.
Recognition of the balance sheet implications of inventory in supply chain management has launched a variety of industry-wide inventory reduction initiatives. Despite all these initiatives to reduce inventory in the supply chain, inventory levels for most companies have remained the same or increased. This course is focused on understanding how to efficiently provide the level of inventory that is really necessary for customer service while minimizing the inventory resulting from poor supply chain management.
Upon completion of this course, you will be able to:
Participants receive a course notebook.
None.
This course is part of the Supply Chain Management (SCM) Certificate and can also be used towards the Strategic Sourcing and Supply Management (SSSM) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
The requirement for high levels of customer service, increasing numbers of SKUs and high labor costs have dramatically increased the complexity of warehouse operations. It is no longer sufficient to manage a warehouse based on a simple, arbitrary “ABC” classification of SKUs, which treats all those in a category as if they were identical. Instead, each decision – such as where to store or where to pick product – must be based on careful engineering and economic analysis. Each SKU must identify its own cheapest, fastest path through the warehouse to the customer and then compete with all the other SKUs for the necessary resources. This results in warehouse operations that are finely tuned to patterns of customer orders and maximally efficient. Learn the concepts necessary to address modern warehouse trade-offs between space and time in optimizing and managing your warehouse.
Essential learning for those who are seeking cost reductions through better handling methods. Also valuable for those who must replace, upgrade, or add material handling equipment. The two-day course will include case examples and a guided exercise to ensure mastery of the techniques presented.
Supply chain and logistics consultants, supply chain engineers and analysts, facility engineers, and warehouse supervisors and team leaders
Upon completion of this course, you will be able to:
Participants receive a course notebook.
None.
This course is part of the Supply Chain Management (SCM) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
This workshop provides an introduction to the field of material handling, including systems analysis, equipment selection, and the relationship of material handling to other activities and operations of the industrial plant or warehouse. It is also an excellent refresher course for those who want an update on the latest trends. You will learn how to plan and analyze material handling systems; how to improve material handling operations; and when to apply material handling automation. Key features are case examples and a guided exercise to ensure your mastery of the techniques presented.
Essential learning for those who are seeking cost reductions through better handling methods. Also valuable for those who must replace, upgrade, or add material handling equipment. The two-day course will include case examples and a guided exercise to ensure mastery of the techniques presented.
Material handling and logistics engineers, manufacturing and process engineers, industrial Engineers and systems analysts, production supervisors and team leaders, warehouse supervisors and team leaders, and cell planning and Lean Manufacturing teams
Upon completion of this course, you will be able to:
Participants receive a course notebook.
None.
This course is part of the Supply Chain Management (SCM) Certificate.
On the first day, please check in at least 30 minutes before the class start time.
In recent years, increasing numbers of companies have become aware that the marketplace encompasses the world, not just the country in which they do business. For example, many firms have found that evaluating offshore sourcing alternatives is essential to a well-run, cost effective organization. Alternatively, by developing export markets, firms have highlighted the need for effective supply networks throughout the world. Conversely, companies located in other countries have also broadened their sourcing and marketing considerations geographically-they look toward global supply strategies and operations to provide competitive advantage through efficiency, effectiveness and differentiation.
Today’s supply managers are finding that they need to do much work in terms of conceptualizing, designing, and implementing initiatives that may be effective globally. In addition, supply managers need to understand the risks inherent in sourcing globally and be able to develop mitigation strategies for these risks. This course and the associated case studies, activities and discussions address key issues and topics that are essential to the global aspects of supply and risk management.
Sourcing and procurement managers, supply chain managers, logistics and supply chain planners, leaders/individuals involved in supply chain cost reduction initiatives, individuals needing to expand sourcing knowledge
Upon completion of this course, you will be able to:
COURSE MATERIALS
Required
Provided
COURSE PREREQUISITES
None.
This course is part of the Strategic Sourcing and Supply Management (SSSM) Certificate.
On the first day, please check in at least 30 minutes before the class start time.