Hauptseminar/Master Seminar:
Deep Generative Models WS2017/2018

Prof. Nassir Navab, Dr. Vasileios Belagiannis, Dr. Federico Tombari, Iro Laina, Christian Rupprecht, Leslie Casas, Kristians Diaz

Type: Advanced Seminar Course Module IN0014
Type: Master Seminar Module IN2107
Type: Seminar Module IN4826
SWS: 2+0
ECTS: 5 Credits
Location: MI 03.13.010
Time: 14:00 - 15:30
Course Language: English

Announcements

• 05.02.2018 - The last lecture is on 06.02.2018 at 18:00.
• 19.10.2017 - The slides from the GAN lecture are now online.
• 26.09.2017 - The schedule is online (please check bellow).
• 24.09.2017 - The papers have been now assigned (please check bellow).
• 14-07-2017 The registration in TUM Matching is now open (14.07.2017 - 19.07.2017).
• 12.07.2017 - The paper list is online (please check bellow).
• Preliminary meeting: Room 03.10.11 on Thursday 06.07.2017 at 1pm.

Introduction

• The recent advances of generative models in Deep Learning will be studied in the Seminar Course. Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs) compose the main topic of interest. A list of papers will be presented to understand the fundamentals of GANs, as well as, their variations (e.g. DCGAN, cGAN, wGAN and etc.). There will be a few application-oriented papers, including combinations with VAEs.

Registration

• Preliminary Meeting: Room 03.10.11 on Thursday 06.07.2017 at 1pm.
• The registration was performed though the TUM Matching Platform.

Requirements

In this Master Seminar (formerly Hauptseminar), each student is asked to select one paper from a list. In order to successfully complete the seminar, participants have to fulfill these requirements:

• Presentation: The selected paper is presented to the other participants (20 minutes presentation 10 minutes questions). Use the CAMP templates for PowerPoint camp-tum-jhu-slides.zip, or Latex: CAMP-latex-template.
• Written Report: A document of maximum 8 pages is written and submitted one week after the presentation. Please download and use the llncs2e.zip Latex template.
• Attendance: Participants have to participate actively in all seminar sessions.

The students are required to attend each seminar presentation which will be held during this course. Each presentation is followed by a discussion and everyone is encouraged to actively participate. Part of the discussions outcome will be included in the report too. The report must include all references used and must be written completely in your own words. Copy and paste will not be tolerated. Both report and presentation have to be done in English.

List of Publications

Nr	Authors	Title	Conference	Link	Tutor	Student
1	Goodfellow, Ian, et al.	Generative adversarial networks	NIPS 2014	https://arxiv.org/abs/1406.2661	Vasilis
2	Radford, Alec, et al.	Unsupervised representation learning with deep convolutional generative adversarial networks	ICLR 2016	https://arxiv.org/abs/1511.06434	Iro	Onur E.
3	Mirza, Mehdi, and Simon Osindero	Conditional generative adversarial nets.	arXiv 2014	https://arxiv.org/abs/1411.1784	Federico	Yasaman R.
4	Arjovsky, Martin et al.	Wasserstein gan	arXiv 2017	https://arxiv.org/abs/1701.07875	Vasilis	Alexander R.
5	Nowozin, Sebastian et al.	f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization	NIPS 2016	https://arxiv.org/abs/1606.00709	Leslie	Maximilian M.
6	Larsen, Anders Boesen Lindbo, et al.	Autoencoding beyond pixels using a learned similarity metric	ICML 2016	https://arxiv.org/abs/1512.09300	Kristians	Kishan S.
7	Dosovitskiy, Alexey, and Thomas Brox	Generating images with perceptual similarity metrics based on deep networks	NIPS 2016	https://arxiv.org/abs/1602.02644	Leslie	Chengnan Y.
8	Lample , Guillaume et al.	Fader Networks: Manipulating Images by Sliding Attributes	arXiv 2017	https://arxiv.org/abs/1706.00409	Oliver	Tom D.
9	Donahue, Jeff, et al.	Adversarial feature learning	ICLR 2017	https://arxiv.org/abs/1605.09782	Federico	Mohd Yawar N. S.
10	Dumoulin, Vincent, et al.	Adversarially learned inference	ICML 2017	https://arxiv.org/abs/1606.00704	Iro	Enes T.
11	Reed, Scott, et al.	Generative adversarial text to image synthesis	ICML 2016	https://arxiv.org/abs/1605.05396	Kristians	Mohammadamin B.
12	Zhu, Jun-Yan, et al.	Unpaired image-to-image translation using cycle-consistent adversarial networks	arXiv 2017	https://arxiv.org/abs/1703.10593	Christian	Kashif J.
13	Tzeng, Eric, et al.	Adversarial discriminative domain adaptation	CVPR 2017	https://arxiv.org/abs/1702.05464	Vasilis	Nikolai M.

Schedule

The schedule for the paper presentations will be announced here during the semester. Each student is required to attend all presentations.

Date	Time	Place	Speaker	Material
06.07.2017	13:00-14:00	MI 03.10.11	V. Belagiannis	Preliminary meeting, Slides
17.10.2017	14:00-15:30	MI 03.13.010	V. Belagiannis (lecture)	Introduction to Generative Adversarial Networks (GANs), Slides, Code
07.11.2017	14:00-15:30	MI 03.13.010	Onur E., Yasaman R.	Unsupervised representation learning with deep convolutional generative adversarial networks, Conditional generative adversarial nets
21.11.2017	14:00-15:30	MI 03.13.010	Alexander R., Maximilian M.	Wasserstein gan, f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization
05.12.2017	14:00-15:30	MI 03.13.010	Kishan S., Chengnan Y.	Autoencoding beyond pixels using a learned similarity metric, Generating images with perceptual similarity metrics based on deep networks
19.12.2017	14:00-15:30	MI 03.13.010	Tom D., Mohd Yawar N. S.	Fader Networks: Manipulating Images by Sliding Attributes, Adversarial feature learning
16.01.2018	14:00-15:30	MI 03.13.010	Enes T., Mohammadamin B.	Adversarially learned inference, Generative adversarial text to image synthesis
23.01.2018	14:00-15:30	MI 03.13.010	Kashif J., Nikolai M.	Unpaired image-to-image translation using cycle-consistent adversarial networks, Adversarial discriminative domain adaptation
06.02.2018	18:00-19:30	MI 03.13.010	V. Belagiannis (lecture)	Introduction to Wasserstein GANs