Stress-testing algorithms via Instance Space Analysis

BIO

Kate Smith-Miles is a Melbourne Laureate Professor of Applied Mathematics in the School of Mathematics and Statistics at The University of Melbourne, Australia. She is also Director of a doctoral training centre for Optimisation Technologies, Integrated Methodologies and Applications (OPTIMA, see optima.org.au), and Associate Dean (Enterprise & Innovation) for the Faculty of Science. Kate obtained a B.Sc(Hons) in Mathematics and a Ph.D. in Electrical Engineering, both from The University of Melbourne. She has held Professorships in three disciplines (mathematics, engineering, information technology), and is involved in many interdisciplinary collaborations and industry partnerships. She has published 2 books on neural networks and data mining, and over 280 refereed journal and international conference papers in the areas of neural networks, optimisation, data mining, and various applied mathematics topics. Her awards include the Australian Mathematical Society Medal in 2010 for distinguished research; the EO Tuck Medal from ANZIAM in 2017 for outstanding research and distinguished service; and the Ren Potts Medal for outstanding research in the theory and practice of operations research from the Australian Society for Operations Research in 2019. She is a Fellow of the Australian Academy of Science, a Fellow of the Australian Mathematical Society, and a past-President of the Australian Mathematical Society. She is frequently invited as keynote speaker at leading international conferences, including IFORS, GECCO, and CPAIOR, to discuss her Instance Space Analysis methodology.

ABSTRACT

Instance Space Analysis (ISA) is a recently developed methodology to support objective testing of algorithms. Rather than reporting algorithm performance on average across a chosen set of test problems, as is standard practice, ISA offers a more nuanced understanding via visualisation of the unique strengths and weaknesses of algorithms across different regions of the instance space that may otherwise be hidden on average. It also facilitates objective assessment of any bias in the chosen test instances, and provides guidance about the adequacy of benchmark test suites and the generation of more diverse and comprehensive test instances to span the instance space. This tutorial provides an overview of the ISA methodology, and the online software tools (see matilda.unimelb.edu.au) that are enabling its worldwide adoption in many disciplines. Several case studies from classical operations research problems will be presented to illustrate the methodology and tools, including timetabling, travelling salesman problem, 0-1 knapsack; and applications to machine learning will also be highlighted.

Seaside Planning: Overview and Implementation (2 sessions)

BIO

She is an associate professor at the Faculty of Engineering and Applied Sciences of the Universidad de Los Andes in Chile. She holds a bachelor’s degree in industrial engineering from the Technologic Institute of Morelia in Mexico, a master’s degree in industrial engineering from Arizona State University in USA; a master’s degree in quality systems and productivity and a PhD in engineering sciences from Monterrey Tech in Mexico. Her research areas are logistics and transport of cargo, maritime shipping and port operations, port governance, competitiveness and performance/efficiency, port gender equity and inter-organizational information systems in ports.
She also works in production planning, vehicle routing problems, and agricultural supply chains (harvesting and distribution) with special focus on small producers. She is currently the Secretary of the Mexican Society of Operations Research. She has been working in several applied research projects with ports in Chile with funding from the Regional Government of Valparaíso, the Regional Government of Arica and Tarapacá, and funding from national agencies in Chile such as Corfo. She collaborated in a regional network of ports in Latin America and the Caribbean (leaded by the Economic System of Latin America and the Caribbean SELA and the Development Bank CAF) between 2014-2019. She is currently executing a research grant (2021-2024) from the National Agency of Research and Development in Chile (ANID), in which she is proposing a Decision Support System for container handling operations in a port terminal. Under the scope of this grant, she has signed an MOU with partners such as DP World, Sitrans, the Port Terminal of Arica, and is currently in process to be signed with the Port Authority of San Antonio and the Port Logistics Community of San Antonio (COLSA) in Chile. During the year of 2021, she developed two studies based on surveys with the participation of several ports in Chile such as the port of Valparaiso, San Antonio, Antofagasta, Arica, Mejillones, Coquimbo and Austral, related to port competitiveness and gender equity. In the gender equity study, she worked with the University of Valparaiso in Chile, and the study had the collaboration of institutions such as the National Service of Women and Gender Equity in Chile (SernaMEG), the Economic Commission of Latin America and the Caribbean from United Nations (ECLAC-UN), the Interamerican Commission of Ports (CIP-OEA), and the associations WISTA (Women in Shipping in Transport Association) and CONECTA Logística Chile. She teaches the courses of Operations Management and Logistics for undergraduate students. She is currently serving as part of the Editorial board of the WMU Journal of Maritime Affairs.

BIO

Eduardo Lalla-Ruiz is an assistant professor at HBE in the IEBIS section at the University of Twente (The Netherlands). He mainly works in the field of logistics and optimization, where he studies and researches current and novel optimization techniques to improve operations management and decision-making. He holds a bachelor's degree in industrial engineering, an advanced engineering degree in industrial automation, and a master of advanced studies in computer science and artificial intelligence from the University of La Laguna (Spain). At that same institution, he got his PhD degree and was awarded an extraordinary doctoral distinction given by the University of La Laguna. After finishing his PhD, he was hired as a researcher and lecturer at the Institute of Information Systems at the University of Hamburg (Germany). There, he became a fellow of the prestigious Alexander von Humboldt Foundation along with the research project about managing disturbances in container terminals. In 2018, he joined the University of Twente. He has published several research papers in specialized journals, collaborated in (international) research projects, served as editor and referee in relevant journals, and participated in the organization of sessions and conferences. Recently, he was awarded as a runner-up for the EURO excellence in practice award 2021 as well as the INFORMS meritorious service award in Transportation Science. His research interests lie in the fields of logistics, artificial intelligence, operations research, mathematical programming, and decision support systems.

ABSTRACT

Maritime transport and international shipping are the backbone of global supply chains and therefore have a significant role in international trade. In this context, container terminals are strategic logistics nodes aimed at providing cargo transfer services while enabling the connection of multiple modes of transportation. Resources at terminals such as equipment, personnel, quay and yard space as well as landside access and parking spaces are scarce. For that reason, it is crucial that managers and relevant stakeholders are supported by decision support systems to fully utilize them. This involves efficiently planning, executing, and controlling the different terminal operations such as berthing, quay deployment, yard allocation, etc., in such a way that their realisation meets certain predefined goals, e.g., short waiting time, low costs, service level, among others.

In this tutorial, we will provide participants with an overview of the main planning decisions at container terminals and depict their main features. We will focus on optimization problems arising at the seaside of the terminal, giving more attention to the berth scheduling operations by means of the well-known berth allocation problem. We will review some variants and approaches for that problem, as well as specific features of real operations at port terminals. The second part of this tutorial is essentially practical, we will hands-on implement a berth allocation model in a computational language and solve some instances with free and commercial solvers. Results will be analyzed to discuss the main trade-offs as well as the main challenges in the literature and practice of this problem.

The tutorial is divided into two sessions:

• Session 1 (90 minutes): Seaside planning: overview
• Session 2 (90 minutes): Seaside planning: model implementation (practical session)

Requirements

For the second session, participants should bring their laptops. We expect that participants may have some basic programming skills and some basic knowledge of mathematical modelling (i.e., MILP, ILP, IP). The tutorial is also helpful for those participants interesting in learning about implementing an optimization model. The participants will receive a brief tutorial to install the required software.

Instructors

Eduardo Lalla-Ruiz
Department of Industrial Engineering and Business Information Systems (IEBIS)
University of Twente
Enschede, Netherlands.
Rosa G. González Ramírez
Faculty of Engineering and
Applied Sciences
Universidad de los Andes Chile
Santiago, Chile.

Inventory Problems with Heterogeneous Customers

BIO

Nesim K. Erkip is a professor of Industrial Engineering at Bilkent University, Ankara. He received his M.S. and Ph.D. from Stanford University and B.S. from Middle East Technical University (METU), Ankara. Before joining Bilkent University in 2005, he worked at METU for over 20 years. He held visiting and research positions at Cornell University, Stanford University, UC Berkeley, Eindhoven University of Technology, New York University, and the Technical University of Munich. He received international awards such as Senior Fulbright Scholar in the USA and August-Wilhelm Scheer Visiting Professorship from the Technical University of Munich, as well as national ones. His main research interests are in multi-echelon inventory theory, distribution systems, supply chains and retailing, and applications of OR. His publications can be found in journals such as Management Science, Operations Research, Manufacturing & Service Operations Management, Production and Operations Management, OMEGA, EJOR, IISE Transactions, IJPE and Journal of Scheduling. He is the author or co-author of many reports and book chapters and mentored dozens of M.S. and Ph.D. students. Additionally, he has been involved in numerous consulting projects on supply chain management and other operations management topics with many companies. He held several administrative positions in two different universities. He is a founding member of the Science and Technology Policy Studies Graduate Program at METU. He has been part of several initiatives preparing reports on “Science and Technology Policy” issues in Turkey during the 1990s. He has been involved in strategic planning activities for non-profit private and public higher education institutions and served as a member of the Board of Trustees of TED University for several years.

ABSTRACT

Inventory Theory is considered one of the earliest topics in the operations management/operations research field. One of the standard assumptions in the bulk of the inventory-related research is the homogeneity of the customers. On the other hand, customers in most practical applications may not be homogenous, as they are affected differently by taste, prices, lead time, service quality, return conditions, and other factors. The decision problem we consider is similar to the decisions of the classical inventory theory, with some customer choices available in the environment. We name these problems as inventory problems with heterogeneous customers.

For a single-item environment, one way of differentiating customers is to assume that they belong to different priority classes. Traditionally, demand for a customer class is assumed to be independent of the demand of the others. For inventory systems with multiple items, the classical assumption considers mutually independent demands, whereas, in reality, substitution is possible. However, customer choice models consider the interaction of assortment variety stocking decisions and customer choice decisions. Hence, these models consider the endogeneity of demands for multiple products.

This tutorial aims to review the contributions of the inventory problems with heterogeneous customers. As much as possible, we aim to exclude the vast literature regarding the revenue management topic and concentrate more on an abstract aspect resulting in heterogeneity. Of course, pricing strategy is one of the factors resulting in customer heterogeneity, where customers are price sensitive.

We review mainly three groups of research:

1) Customers are grouped in different priority classes for a single item inventory problem.

2) Customers are grouped in different priority classes for a multi-item inventory problem under substitution.

3) Customers are differentiated by their perception of the available choices, reacting to decisions made by the solver of the inventory problem.

The focal point of the tutorial is to emphasize the role of considering demand endogeneity for decision-making in inventory systems.

Machine Learning

BIO

Andrés Gómez received his B.S. in Mathematics and B.S. in Computer Science from the Universidad de los Andes (Colombia) , and he then obtained his M.S. and Ph.D. in Industrial Engineering and Operations Research from the University of California Berkeley. From 2017 to 2019, Dr. Gómez worked as an Assistant Professor in the Department of Industrial Engineering at the University of Pittsburgh, and since 2019 he is an Assistant Professor in the Department of Industrial and Systems Engineering at the University of Southern California. Dr. Gómez research focuses on developing new theory and tools for challenging mixed-integer (nonlinear) optimization problems arising in statistical learning and operations research. His research is funded by multiple grants and gifts from the National Science Foundation, the US Air Force Office of Scientific Research and Google Research, among others.

ABSTRACT

Machine learning (ML) and artificial intelligence (AI) methods have advanced tremendously over the past three decades: accurate facial/voice recognition technologies are now commonplace, and AIs consistently exhibit super-human intelligence in games they are trained to play (such as Go and Poker). However, most ML/AI methods require two settings to be effective: access to massive amounts of data, and relatively low-stakes situations. Moreover, they are extremely sensitive to data quality, and tend to be biased. As a consequence, AI/ML is not effective or reliable in the context of several important engineering and societal problems.

There is a recent stream of results showcasing that mixed-integer optimization (MIO) can be used to improve current ML/AI methods. In particular, MIO can be used to incorporate a variety of prior beliefs on the structure of the process of interest (such as sparsity) in the forms of logical priors, be used to design interpretable models, can easily accommodate fairness considerations and handle robust variants of common ML methods. In this tutorial, we review recent applications of MIO methods to improve AI/ML models. We explore the current capabilities and limitations of MIO solvers, and discuss the theory and design of effective MIO techniques.

Quantum Computing and Combinatorial Optimization

BIO

Dr. Carleton Coffrin is a senior scientist at Los Alamos National Laboratory with expertise in computer science, optimization algorithms, and artificial intelligence. Dr. Coffrin started exploring the field of quantum computation in 2016 and has had the opportunity to benchmark multiple generations of quantum computing hardware from vendors including D-Wave Systems and IBM. Dr. Coffrin was one of the founders of LANL's Quantum Computing Summer School program serving as a co-lead for three years (2018, 2019, 2020) and is a leading contributor to a Quantum Economic Development Consortium (QED-C) effort to develop optimization benchmarks for quantum computing platforms. In addition to quantum computing research, Dr. Coffrin has expertise in benchmarking optimization methods for energy systems applications, especially the AC Optimal Power Flow problem, and is a core contributor to the design and operation of ARPA-e’s Grid Optimization Competition.

BIO

Fred Glover is Chief Scientific Officer of Entanglement, Inc., USA, in charge of algorithmic design and strategic planning initiatives. He also holds the title of Distinguished University Professor, Emeritus, in the School of Engineering, the Department of Applied Mathematics and in the Leeds School of Business at the University of Colorado. He has authored or co-authored more than 500 published articles and eight books in the fields of mathematical optimization, computer science and artificial intelligence, and is the originator of “Tabu Search,” an adaptive memory programming algorithm for mathematical optimization in complex search spaces, for which Google returns more than 1,000,000 results.

Dr. Glover is an elected member of the U. S. National Academy of Engineering and is the recipient of the von Neumann Theory Prize, the highest honor of the Institute of Operations Research and Management Science. His numerous other awards and honorary fellowships include those from the Institute of Electrical and Electronics Engineers (IEEE), the American Association for the Advancement of Science (AAAS), the NATO Division of Scientific Affairs, the Institute of Operations Research and Management Science (INFORMS), the Decision Sciences Institute (DSI), the U.S. Defense Information Systems Agency (DISA), the National Renewable Energy Laboratory (NREL), the American Assembly of Collegiate Schools of Business (AACSB), Alpha Iota Delta, the Glushkov Institute of Cybernetics of the Ukrainian Academy of Science, and the Miller Institute for Basic Research in Science. He also serves on advisory boards for numerous journals and professional organizations, and has co-founded the companies Analysis, Research and Computation, Inc. (now within Science Applications, Inc.), Heuristec, Inc. (now within Tomax, Inc.) and OptTek Systems, Inc.

BIO

Dr. Gary A. Kochenberger has a BS in electrical engineering and a PhD in Management Science from the University of Colorado. He is currently the Chief Optimization Officer at Entanglement, Inc. His research interests include combinatorial optimization, resource allocation, pattern classification, data mining, and related areas. He has held editorial positions at several major journals and is currently on the editorial board of the journal Networks. He has published 4 books and more than 100 articles on operations research and optimization.

ABSTRACT

Recent years have witnessed a series of exciting developments in quantum computing and its potential for solving combinatorial optimization problems. In this two-part talk, we first give an overview of quantum computing technology, its present state, breaking developments, and most promising forms. This introduction to the technology is followed by an overview of how quantum-based solvers are being used to solve combinatorial optimization problems, including an introduction to the fundamental QUBO model, how it is formed and how it is being used to solve important problems. Accompanying this, we describe Important extensions and variations of the QUBO model that are expanding its range of application, bolstered by several explicit examples and substantial computational experience.

Authors:
Carleton Coffrin, Staff Scientist, National Laboratory at Los Alamos
Fred Glover, Chief Scientific Officer, Entanglement, Inc.
Gary Kochenberger, Chief Optimization Officer, Entanglement, Inc