General Mathematics Colloquium
Speaker: Tim van Erven (KdVI)
Event details from: General Mathematics Colloquium
- Date
- 26 November 2020
- Time
- 16:00 – 16:45
Title
Online Convex Optimization: From Proving Minimax Theorems to Training Machine Learning Models by Playing Repeated Games
Abstract:
Modeling sequential decision problems as repeated games has a long history in statistics, information theory and machine learning. Using the modern unified setting of online convex optimization, I will introduce several canonical results and highlight applications that range from an elementary proof of von Neumann's minimax theorem to optimization of (deep) learning models in machine learning.
Location:
Online