What Is Stein’s method for Distributional Approximation?
- Event Type
- Seminar/Symposium
- Sponsor
- AMS Graduate Chapter
- Location
- Davenport 169
- Date
- Dec 2, 2025 4:00 - 4:50 pm
- Speaker
- Partha Dey
- Contact
- David Gallardo
- davidg7@illinois.edu
- Views
- 11
- Originating Calendar
- What is... Seminars
An installment of the What Is... Seminar Series. Snacks, tea and coffee will provided beforehand at Harker 300 at 3:30pm.
Speaker: Partha Dey
Title: What Is Stein’s method for Distributional Approximation
Abstract Provided by Partha Dey: Stein's method is a powerful probabilistic tool for approximating probability distributions and quantifying the error in these approximations. Introduced by Charles Stein in the 70s, this method provides a generic way to measure the distance between a general and a target distribution. This method is constructive, provides an explicit error bound, and works well for dependent random variables and a vast class of target distributions. We'll go over the underlying philosophy and give a few examples related to Normal and Poisson approximation, and Machine Learning.