WiSe 2018/19

A deeper look into deep learning – perspectives from statistics and information theory

Univ.-Prof. Dr. Michael Wand

Shortname: 08.079.630
Course No.: 08.079.630

Additional information

The seminar itself will be held during the lecture break after the winter term. The introductory lecture will be held during the winter semester lecture time. Concrete times will be decided in the prepatory meeting:

Important: Please make sure to attend the prepatory meeting

Tuesday, Oktober 23 2018, 16:15-17:00h in Room 03-424 (Staudingerweg 9)

If you cannot attend this first meeting, please send an email to the lecturer (Michael.Wand (at) uni-m...)

Requirements / organisational issues

Basic knowledge in machine learning or computer vision is very useful.

The seminar will be accompanied by a short introductiory lecture to information theory, statistical tools, optimization and basic deep learning techniques (roughly 1/2 of a semester) so that the seminar can in principle be attended without prior attendence of any advanced courses.

Contents

Will will discuss papers that aim at explaining why deep learning works at all (something we do not really understand at this point). This will also include interdisziplinary approaches, for example from statistical physics.
Semester: WiSe 2018/19