Pierre Nyquist: A series of mini-talks on large deviations, stochastic simulation and their roles in the mathematics of data and AI
Time: Tue 2021-05-04 11.15
Lecturer: Pierre Nyquist (KTH)
Location: Zoom, meeting ID: 625 8662 8413
This will not be a traditional talk where I start with a well-defined research question and end with some kind of answer. Instead, it will be series of “the first 10–15 minutes” of several talks on large deviations and stochastic simulation. These are topics that are ubiquitous in a variety of areas within probability and statistics, or even more broadly in applied mathematics, but have yet to be used extensively in the area of data science. I will start with a brief introduction to the theory of large deviations, one of the cornerstones of modern probability, and then discuss problems involving e.g. gradient flows for interacting particle systems and Markov chain Monte Carlo methods for sampling from complex distributions, and how these link to important questions in data science. If time permits I will end the talk with an outline of some ongoing and future work on the mathematical foundation of data science. As we go along, I will refer back to previous talks in the seminar series to highlight how the topics I discuss have already come up in different contexts. No previous knowledge of any of the topics mentioned is required.