Hauptseminar: Deep Learning

Talks

1. Ashray Jain Deep learning basics April 24		Slides
2. Lecture by Prof. Rosenow Convolutional neural networks May 08
3. Felix Stündel Recurrent neural networks May 15
4. Johannes Ewald Graph neural networks May 22
5. Yoon Thelge Autoencoders May 29		Slides		Code
6. Hamid Al Zubaidi Generative models June 05
7. Stephan Meyer Transfomers June 12

Written Assignments

Course Information

Time of talks:	Wednesdays, 17:15 R114, ITP, Brüderstraße 16 (Talks)
First meeting:	April 10, 2024, 17:15, R114, ITP, Brüderstraße 16 (General information)

Topic	References
1. Deep learning basics (building blocks of NNs, training, fully connected networks)	Ref. [1] (Sec. 3,4,7) Ref. [2] (Sec. 9)
2. Convolutional neural networks	Ref. [1] (Sec. 8) Ref. [2] (Sec. 10)
3. Recurrent neural networks	Ref. [1] (Sec. 9)
4. Graph networks	Ref. [1] (Sec. 10)
5. Interpretability of networks	Ref. [1] (Sec. 12)
6. Objective functions	Ref. [1] (Sec. 14)
7. Unsupervised learning	Ref. [2] (Sec. 12, 13)
8. Autoencoders	Ref. [1] (Sec. 17) Ref. [2] (Sec. 17)
9. Generative models	Ref. [1] (Sec. 18) Ref. [2] (Sec. 15, 16)
10. Transformers, "Attention is all you need"	Ref. [3]
References

[1]	Deep learning for physics research, Martin Erdmann, Jonas Glombitza, Gregor Kasieczka, and Uwe Klemradt, DOI 10.1142/12294
[2]	Physics Reports 810 1–124 (2019), A high-bias, low-variance introduction to Machine Learning for physicists
[3]	Advances in neural information processing systems 30 (2017), Attention is all you need., Vaswani, Ashish, et al.