Personal tools
You are here: Home CG seminar 2021 Some methods for reasoning about topology & global geometry in machine learning & robotics
« December 2020 »
December
SuMoTuWeThFrSa
12345
6789101112
13141516171819
20212223242526
2728293031
Log in


Forgot your password?
 

Some methods for reasoning about topology & global geometry in machine learning & robotics

Wednesday, November 25th, 16:10

underline

Florian Pokorny, KTH Royal Institute of Technology

Abstract:

In this talk, I will discuss some applications of ideas from topological
data analysis based on the use of Delaunay-Cech and Alpha complexes.
Use cases in robotics include motion clustering as well as applications
to motion planning, caging/path non-existence and the analysis of
configuration spaces, while applications to machine learning include the
classification and topological data analysis of image data. In
particular, I will discuss some or our recent work on developing
randomized approximation algorithms that allow one to apply these ideas
in higher dimensions [ICML'19 & KDD'20, Polianskii & Pokorny] and we
will then discuss some of the challenges that remain in this area.
 
Document Actions