250063 VO Mathematics of Deep Learning (2024S)
Labels
ON-SITE
Registration/Deregistration
Note: The time of your registration within the registration period has no effect on the allocation of places (no first come, first served).
Details
Language: English
Lecturers
Classes (iCal) - next class is marked with N
Monday
04.03.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
07.03.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
11.03.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
14.03.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
18.03.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
21.03.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
08.04.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
11.04.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
15.04.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
18.04.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
22.04.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
25.04.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
29.04.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
02.05.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
N
Monday
06.05.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Monday
13.05.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
16.05.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Thursday
23.05.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
27.05.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Monday
03.06.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
06.06.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
10.06.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
13.06.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
17.06.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
20.06.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Monday
24.06.
13:15 - 14:45
Seminarraum 16 Oskar-Morgenstern-Platz 1 3.Stock
Thursday
27.06.
13:15 - 14:45
Seminarraum 8 Oskar-Morgenstern-Platz 1 2.Stock
Information
Aims, contents and method of the course
Assessment and permitted materials
Depending on the number of students that take this class, we will have either a written exam or an oral exam at the end of the lecture.
Minimum requirements and assessment criteria
To pass this class a student needs to demonstrate that they have understood the material of the course on a basic level. For the best grade, a thorough understanding of each result is necessary.
Examination topics
Everything that was said in the lecture.
Reading list
There will be course notes that will be published throughout the course.
Association in the course directory
MAMV; MANV; MSTV;
Last modified: We 28.02.2024 10:26
2. Feed-forward neural networks: The main building block of deep learning is that of a neural network. We will introduce this in a formal way.
3. Universal Approximation: We will study multiple results that study the (absence of) limitations of deep neural networks to represent general functions.
4. Connection to Splines: There are close relationships between neural networks and classical approximation methods. We will discuss these.
5. ReLU neural networks: There is a special subclass of neural networks that is the most frequently used. We study this special case in more detail.
6. Affine pieces of ReLU neural networks: A standard tool to understand what deep neural networks can and cannot do is to count the number of affine regions that they can generate. We will find upper and lower bounds for this.
7. Deep ReLU neural networks: We study the effect of depth, specifically for deep ReLU neural networks and find reasons for the fact that in practice deeper architectures reign supreme.
8. Curse of Dimensionality: We study the phenomenon of high dimensional approximation, which neural networks seemingly do well, despite the fact that this is typically very hard.
9. Interpolation: Deep neural networks can interpolate data under certain assumptions. We will formalize those.
10. Training of deep neural networks: We will understand how to train neural networks, and under which conditions this training can work
11. Loss landscape analysis: The optimization problem can be understood by studying the topography of the so-called loss landscape.
12. Neural network spaces: The set of all neural networks with a fixed architecture will be studied and conclusions for optimization will be drawn.
13. Generalization: We will describe general statistical learning theory in the context of deep neural networks. This will show under which conditions one can generalize from training sets to test sets.
14. Generalization in the overparameterized regime: The most common regime, i.e., overparameterized is excluded from the results before. We will describe why in practice things still work
15. Adversarial examples and Robustness: Finally, we study under which conditions deep neural networks are robust to changes in their inputs.