ODSL Forum: Introduction to Topological Machine Learning

by Prof. Bastian Rieck (Helmholtz Munich and TUM)

Friday 28 Apr 2023, 14:00 → 15:00 Europe/Berlin

Basement Seminar Room (Origins Building)

Topological concepts, with their focus on *connectivity* of data, 
have received increased attention over the past years. This is 
primarily because of their robustness guarantees in the presence of 
noise, and their flexibility in terms of metrics or similarity 
measures. In this talk, I want to provide an overview over the 
nascent field of *topological machine learning*, which uses 
topological concepts in the form of additional inductive biases, 
regularisation terms, or even new layers for deep learning 
architectures. As a running example, I will talk about preliminary 
work that analyses point clouds arising from particle collisions, 
thus showcasing the potential for topological machine learning 
methods in physics.