A General Framework for Analyzing Stochastic Dynamics in Learning AlgorithmsSpeaker: Chi-Ning
Title: A General Framework for Analyzing Stochastic Dynamics in Learning Algorithms
Date: 04 Mar 2021 17:00-18:00 EST
Abstract: Stochastic process is a fundamental and ubiquitous mathematical object, however, it is relatively less used and studied in CS due to its continuous nature. In this talk, we are going to present a general framework for analyzing the stochastic processes that appear in many applications on CS problems. Our framework composes standard techniques from probability theory to give a streamlined three-step recipe with a general and flexible principle to tackle the “chicken and egg” problem.
The talk assumes no background in advanced probability theory. I will walk the audience through important concepts such as filtration, adapted stochastic process, and stopping time using intuitive language and examples. Hopefully after the talk you will be able to have a new analysis tool for your future research!
This is a joint work with Juspreet Singh Sandhu, Brabeeba Wang, and Tiancheng Yu. Link to the paper.