36-745 Game-Theoretic Statistical Inference: Betting, E-Values, and Martingales

12 units, Carnegie Mellon University

We will study the gametheoretic foundation for statistics and its application to prediction, testing, and estimation. It will distinctively employ modern concepts around betting, evalues, and martingales, effectively bridging Bayesian, frequentist, and modelfree or adversarial perspectives on these topics. This course is taught remotely as a cooperative course with Rutgers University and the University of Waterloo. All three professors will rotate lectures to students at all three universities via Zoom. Students from the three universities will be present on the same Zoom the students will register through and be assessed and graded by their respective universities. The prerequisites are basic graduate level or advanced undergraduate training in statistics or probability.

Prerequisites: none listed

Courseweave needs JavaScript to show sections, workloads and planning tools.