Tutorial Sessions

Speaker: Professor Ben Recht

Bio: Benjamin Recht is a Professor in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. He was previously an Assistant Professor in the Department of Computer Sciences at the University of Wisconsin-Madison. Ben received his B.S. in Mathematics from the University of Chicago, and received a M.S. and PhD from the MIT Media Laboratory. After completing his doctoral work, he was a postdoctoral fellow in the Center for the Mathematics of Information at Caltech.

Ben is the recipient of a Presidential Early Career Award for Scientists and Engineers, an Alfred P. Sloan Research Fellowship, the 2012 SIAM/MOS Lagrange Prize in Continuous Optimization, the 2014 Jamon Prize, the 2015 William O. Baker Award for Initiatives in Research, and the 2017 and 2020 NeurIPS Test of Time Awards. He has served on the Editorial Boards of the Journal for Machine Learning Research and Mathematical Programming. He also cofounded the Conference on Learning for Decision and Control.

Title: Statistics when n equals 1

Abstract: Statistical decision-making draws general inferences about people from studies of populations, but such population inferences tell us little about what to do with any particular person. In this tutorial, I’ll try to answer the apparently oxymoronic question of what it means to do statistics when n equals 1. Throughout, I aim to highlight the role of mathematical and qualitative theory in our understanding, treating, and optimizing individual outcomes.

I’ll first present a review of the basics of statistical decision making based on actuarial methods and randomized trials from the perspective of signal processing. From this statistical foundation, I’ll turn to individuals. With a focus on clinical practice, I’ll review the systems-level view of health based on Walter Cannon’s notion of homeostasis. Cannon conceived of the body as a control system—a complex dynamical system actively working to maintain itself in a stable state despite adversarial engagement with an uncertain and hazardous environment. I’ll present Cannon’s concepts through a contemporary lens, drawing on ideas from feedback control that illuminate the necessary architectures for homeostasis. Identifying these patterns can guide positive interventions that can steer dysregulated systems back to stable behavior. That is, a homeostatic view can help us envision experimentation where an individual is simultaneously the treatment and the control group.

Speaker: Professor Ramesh Johari

Bio: Ramesh Johari is a Professor at Stanford University, with a full-time appointment in the Department of Management Science and Engineering (MS&E), and a courtesy appointment in the Department of Electrical Engineering (EE). He is an associate director of Stanford Data Science. He is a member of the Operations Research group and the Social Algorithms Lab (SOAL) in MS&E, the Information Systems Laboratory in EE, and the Institute for Computational and Mathematical Engineering. He received an A.B. in Mathematics from Harvard, a Certificate of Advanced Study in Mathematics from Cambridge, and a Ph.D. in Electrical Engineering and Computer Science from MIT.

He is the recipient of a British Marshall Scholarship, First Place in the INFORMS George E. Nicholson Student Paper Competition, the George M. Sprowls Award for the best doctoral thesis in computer science at MIT, Honorable Mention for the ACM Doctoral Dissertation Award, the Okawa Foundation Research Grant, the MS&E Graduate Teaching Award, the INFORMS Telecommunications Section Doctoral Dissertation Award, the NSF CAREER Award, and the Cisco Faculty Scholarship. He has served on the program committees of ACM Economics and Computation (including co-chairing the conference in 2019), ACM SIGCOMM, IEEE Infocom, and ACM SIGMETRICS, as the track chair for the Internet Economics and Monetization Track at WWW, and as a founding co-organizer of the Marketplace Innovation Workshop. He served as founding Area Co-Editor of the Revenue Management and Market Analytics Area for Operations Research, and as associate editor for Management Science (in the Stochastic Models and Simulation area) and Stochastic Systems.

Title: Experimentation, Interference, and Capacity Constraints: Recent Results and Future Directions

Abstract: Two-sided marketplace platforms often run experiments to test the effect of an intervention before launching it platform-wide. However, estimates of the treatment effect obtained in these experiments can be biased, due to interference arising from marketplace competition: e.g., when buyers are randomized to treatment and control, they compete for the same limited supply.  The resulting biases can impact the platform’s ability to make data-driven decisions.

In this tutorial, we survey recent results on this problem.  A central theme of the tutorial will be to illustrate that the “state” of the system as measured by available capacity (i.e., inventory) provides a natural modeling device to study the impact of interference.  This observation allows us to naturally study experimental design and interference using the tools of stochastic modeling.  The talk will use this approach to describe the impact of interference on estimation bias and subsequent decision-making; examples and applications; some proposed solutions; and open directions for future work.