Course Information
This class is fully booked.
Description
This seminar is designed for master students of Scientific Computing and Mathematics. We will address Machine Learning from a mathematical point of view and discuss its relations to well-known mathematical concepts. This semester, we restart our seminar series and focus on different network architectures. We explore different Deep Neural Networks and their relation to Ordinary Differential Equations.
Note that master students will be prioritized, but if the class does not fill up with master students, late bachelor students will be considered for vacant spots in the class.
Timeslot and Location
Our recurring timeslot will be Thursdays 16:00h sharp(!) with meetings expected to run about 60-75 minutes long. We will be meeting in Seminarroom 2 of INF 205 (Mathematikon).
Dates and Topics
Topics have been assigned to all participants. The list of topics presented at the corresponding dates and their respective resources are:
- 2024-10-31 Introduction Machine Learning I (chap. 5 of GBC2016) (LB)
- 2024-11-07 Introduction Machine Learning II (chap. 5 of GBC2016) (AB)
- 2024-11-14 Feed Forward Network (chap. 6 of GBC2016, chap. 7.2 HF2018) (DB)
- 2024-11-21 ResNet ( HZRS2016 - 240k citations) (WJ)
- 2024-12-12 Convolutional Network (chap. 9 of GBC2016, chap. 7.2.3 of HF2018) (YC)
- 2024-12-19 Recurrent Neural Network (chap. 10 of GBC2016, chap. 8.3.1 of HF2018 (BL)
- 2025-01-09 Long Short Term Memory Network (chap. 10.10 of GBC2016, HS1997 (113 k citations)) (LK)
- 2025-01-16 Auto Encoder (chap. 14 of GBC2016) (QC)
- 2025-01-23 Transformer (double feature topic for 2 people) ( VSP2017) (FV, QD)
Presentations and Final Grade
For a successful completion of the seminar, students will be asked to give a presentation of about 45 minutes and submit a corresponding write-up. The final grade will be made up of the individual grades of both components.
Submit your slides and write-up at least 24 hours before your presentations via E-Mail .
Resources
- Goodfellow, Bengio, Courville: Deep Learning (GBC2016)
- Hoogendoorn, Funk: Machine Learning for the Quantified Self (HF2018)
- He, Zhang, Ren, Sun: Deep residual learning for image recognition (HZRS2016)
- Hochreiter, Schmidhuber: Long Short-Term Memory (HS1997)
- Vaswani, Shazeer, Parmar, Uszkoreit, Jones, Gomez, Kaiser, Polosukhin: Attention is all you need (VSP2017)