Chancellor and Ford Foundation Professor of Engineering
Operations Research and Public Policy: Making a Difference
Our nation and the world have undergone an enormous number of societal changes in 2020 that will impact our lives well into the future. This has created an even greater need for data-driven analysis and modeling to address a growing list of new challenges. Operations research has a long history of impacting public policy, in such diverse areas as the military, homeland security, government, and public health. Several of these issues have been central themes throughout my academic career, including risk-based aviation security, election forecasting, and computational redistricting. This talk discusses a number of these research problems, focusing on how operations research has made a difference, and how the operations research community can provide valuable service to our nation by using our expertise to address emerging opportunities in this new world.
Founder Professor of Computer Science
University of Illinois at Urbana-Champaign
The Future of Federal Statistics and the Role of the Chief Statistician
Technology and the availability of data has significantly changed inputs to federal statistical data. Users want reliable, timely, and more granular data but statistical agencies face ongoing budget limitations as they attempt to modernize. Maintaining data quality while achieving efficiencies in data collection, such as those built into the 2020 Census field operations, presents many challenges The Chief Statistician of the US is charged with coordinating the federal statistical system, setting quality standards, and leading change. Explore what change might look like with the former Chief Statistician, who presents a vision of how far we have come recently and what lies ahead.
former Deputy Director and COO
U.S. Census Bureau
Stop Chasing Unicorns and Start Hiring the Data Scientists You Need
We’re almost a decade into the data science hiring frenzy, but despite the plethora of quantitative degree programs and bootcamps popping up to train professionals flooding into the field – the demand for analytics talent has continued to greatly outpace the supply. And, as the field continues to evolve, your recruiting strategies must adjust to new shifts in the market, including industry disruption and work from home strategies due to the COVID-19 crisis. What motivates quantitative professionals in today’s market? How much do they earn? How can you secure the talent you need? This session contains Burtch Works’ latest research on today’s analytical talent and how the analytics hiring market has changed over the past year, as well as actionable insights from data science recruiting expert, Linda Burtch, on how to use this data to adapt your recruiting strategy for the today’s market.
Founder & Managing Director
Mathematical Optimization for Social Distancing
The spread of viruses such as SARS-CoV-2 brought new challenges to our society, including a stronger focus on safety across all businesses. In particular, many countries have imposed a minimum social distance between people in order to ensure their safety. This brings new challenges to many customer-related businesses, such as restaurants, offices, etc., on how to located their facilities under distancing constraints. In this talk we propose a parallelism between this problem and the one of locating wind turbines in an offshore area. Even if the two problems may seems very different, there are many analogies between them. In particular, both problems require fitting facilities (turbines or customers) in a given area while ensuring a minimum distance between them. Similarly to nearby customers who can infect each other, also nearby turbines “infect” each other by casting wind shadows (the so-called “wake effect”) that cause production losses. In both problems we want to minimize the overall interference/infection, hence optimal solutions will favor layouts where facilities are as spread as possible. The discovery of this parallelism between the two applications allowed us to apply Mathematical Optimization algorithms originally designed for wind farms, to produce optimized facility layouts subject to social distancing constraints as those arising in the time of COVID-19 pandemic. These methods allow us to challenge the current (manual) layouts and provide new insights on how to improve them. In particular we show that optimized layouts are far from trivial to design and that Mathematical Optimization can make an impact, helping businesses while ensuring safety.
IFORS Distinguished Lecturer
Professor of Operation Research at the Department of Information Engineering
University of Padua
Quantum Computing – Why, What, When, and How
Head of Business Development
Optimization for Machine Learning: Insights and Challenges
What is the mathematical optimization viewpoint on machine learning, and how does it scale to modern applications?
The origins of linear programming stem from military resource allocation over tens of variables and constraints. Recently, techniques from mathematical optimization were used to optimize 175 billion parameters of a highly non-linear language model.
In this talk we’ll survey the algorithms arising from early developments in optimization theory, to giga-scale modern problems that lie at the heart of artificial intelligence research. We’ll describe some recent developments, insights and challenges facing researchers in our field.
Professor of Computer Science
Chief Analytics Officer
Rebooting Simulation for Big Data, Big Computing and Big Consequences
Lurking in a track or two at the INFORMS Annual Meeting is the “simulation” crowd. Are they still generating random numbers, reducing variance and writing code? Like many fields of study, simulation has been greatly influenced by its history, particularly Conway’s 1963 Management Science paper on “tactical problems in digital simulation” and early languages like GPSS. However, the following statements, none of which were true in 1963, are hard to dispute: (a) There exists, and we can store, lots of data; (b) parallel computing capacity can be rented, and is cheap; and (c) critical societal decisions are based on large-scale computer models. This talk argues that these facts compel a reboot of computer simulation thinking in operations research and management science, and will explore the consequences.
Walter P. Murphy Professor of the Department of Industrial Engineering and Management Sciences
Distinguished Visiting Scholar
From Patient to Population: Integrating Personalized Medicine and Public Health
While personalized, or precision, medicine deals with individuals, the choices and behaivors of these individuals affect population level health. Public policy makers need to consider individual behaivor when proposing interventions. This has become painfully obvious with the current pandemic, and if results are available I will discuss the impact of compliance with quarantine, travel recommendations, and face coverings on the spread of the SARS-CoV-2. In addition, the individual plays an important role in a much broader range of health policy problems. I will discuss earlier work related to smoking cessation and colorectal cancer screening. In these cases, patients have choices not only in treatment/screening type but also in adherence to treatment. By considering patient choice and behavior, we can not only better assess the true effectiveness of interventions but also design more targeted interventions. Lastly, our recommendations regarding who should be targeted for treatment impacts system resources, and ultimately the broader patient population as resources become scarce.
Professor of Personalized Medicine in the Edward P. Fitts Department of Industrial and Systems Engineering
North Carolina State University
Evolution of Retail Supply Chains – A Practitioners Perspective
Retail supply chains a few decades ago were about getting products from a local manufacturer to a local seller. The focus was on physical flow. It then evolved to a global scale where the winners extracted value by focusing on scale and efficiency. This is where the physical flow combined with financial angle gained significance. Then the focus shifted to efficiency with real-time visibility and control. Information flow took its place along with financial and physical flow. Still in the retail world, supply chain teams played a secondary role; merchants and store operations organizations ran the show. With the evolution of eCommerce, supply chain is taking its place in the boardroom. Now supply chain is defining the flow of retailer’s strategy; on how to balance cost with service, how to provide innovative options to the shoppers and how to get the right product to the right customer at the right time. What does that mean for the retail industry, what does it mean for operations research and data science practitioners?
Vice President: Supply Chain Strategy, Planning, and Execution
American Eagle Outfitters
Statistical Learning in Operations: The Interplay between Online and Offline Learning
Traditionally, statistical learning is focused on either (i) online learning where data is generated online according to some unknown model; or (ii) offline learning where the entire data is available at the beginning of the process. In this talk we show that combining both approaches can accelerate learning. First, we show the impact of pre-existing offline data on online learning and characterize conditions under which offline data helps (does not help) improve online learning. Second, we show how difficult online learning problems can be reduced to well-understood offline regression problems. We demonstrate the impact of our work in the context of dynamic pricing.
Professor of Engineering Systems
AI is the Right Term for Our INFORMS Profession
Ever since I went to grad school in the mid 90s, I’ve seen how this profession has had trouble describing what it is we studied and what we do. Around 2011 (or so), we slowly tried to fit into the term Analytics.
I’m now convinced we should proudly say that the INFORMS profession is part of the AI movement– we should say that we do AI.
In this talk, I’ll explain how we came to embrace the term Analytics, how the term AI has evolved, and why we can now lay claim to using AI. I’ll even give my opinion on why the term AI is better than the term Analytics for us.
A term like AI will never be clean. In this talk, I’ll try to give the right nuances and caveats. My goal is that you can explain your involvement with AI to someone at a cocktail party. Also, my goal is to explain it so that if that person at the cocktail party works on self-driving cars or works for OpenAI, you can confidently explain how you fit in to their definition of AI.
I might not convince you, but hopefully I’ll help you shape your own definition, and you’ll better understand how others are using the term.