Seminar: Ausgewählte Themen des Machine Learning (MA)
Overall Information
Organizer: Prof. Dr. Andreas Hotho, Tobias Koopmann, Janna Omeliyanenko
Application: lsx-seminar[at]informatik.uni-wuerzburg.de (please include your matriculation number and your graduate program)
For questions or to reserve a topic/places please write an email to lsx-seminar[at]informatik.uni-wuerzburg.de. Reservations for places in our seminar are already open!
Please always include your study program and matriculation number and whether you have already heard related courses such as data mining or have other prior knowledge.
We will invite students in the WueCampus course after the preliminary meeting.
Preliminary meeting:
Time: 21.10.2021 at 10.00 AM
Location: Zoom ( https://uni-wuerzburg.zoom.us/j/99445101784?pwd=YlBHbStqSXhrMmprN2J6UElHTjQwQT09 )
or via ID: 994 4510 1784 and password: 783952
Corona News
Due to the current COVID-19 situation, the seminar will be held online.
At the initial meeting, we will finally assign topics to students. Topics or reservations can be reserved beforehand (see above).
All meetings and lectures will be hosted via ZOOM as per current status. We ask that you have a working camera.
ZOOM
ZOOM requires a pre-installed client, which is available at https://zoom.us/download. However, an account is not required to participate. ZOOM is also available on Android and IOS.
(ZOOM also works in the browser. However, we cannot recommend this variant due to poorer performance).
Topic assignment and preliminary discussion
Topics can be reserved before the preliminary meeting date. The topics will be announced here on this page some time before the preliminary meeting.
At the preliminary meeting, the topics - if not already assigned - will be distributed to the seminar participants present. Attendance is compulsory. Participants who are already working on a topic but are absent without prior agreement lose the right to their topic/place.
Topics
The basis for the presentations are scientific papers. Each paper listed here corresponds to one topic.
At the first meeting, remaining topics or topics that have become free again will be distributed to students who are present. Those who wish to do so can also be put on a waiting list (attendance is still required).
Reserved Topics
-
Hypergraph Neural Networks. . In AAAI, pp. 3558–3565. AAAI Press, 2019.
- [ BibTeX ]
- [ URL ]
- [ BibSonomy-Post ]
-
Modeling Relational Data with Graph Convolutional Networks. . In ESWC, Vol. 10843 of Lecture Notes in Computer Science, A. Gangemi, R. Navigli, M.-E. Vidal, P. Hitzler, R. Troncy, L. Hollink, A. Tordai, M. Alam (eds.), pp. 593–607. Springer, 2018.
- [ BibTeX ]
- [ URL ]
- [ BibSonomy-Post ]
-
Inductive Representation Learning on Large Graphs. . In NIPS, I. Guyon, U. von Luxburg, S. Bengio, H. M. Wallach, R. Fergus, S. V. N. Vishwanathan, R. Garnett (eds.), pp. 1024–1034. 2017.
- [ BibTeX ]
- [ URL ]
- [ BibSonomy-Post ]
-
Semi-supervised classification with graph convolutional networks. . In arXiv preprint arXiv:1609.02907. 2016.
- [ BibTeX ]
- [ BibSonomy-Post ]
Proof of performance
At the end of the deadlines specified above, each must be submitted to the supervisor via email as a PDF:
- the slide set for the presentation
- a 4-6 - page LaTeX elaboration in the format described here (double column, including bibliography)
Prior consultation with the supervisor is expressly desired. In addition to the bibliography of the paper, all references used are to be documented in http://www.bibsonomy.org, and tagged with the tags of the respective topic and other meaningful tags in the corresponding group.
The duration of the presentation is mandatory 20 minutes, after the presentation there will be a discussion in the seminar group (approx. 10 minutes). The presentation (incl. slides and abstract) will be graded with 40%, the paper with 30%. The last 30% evaluates the preparation, participation in the discussion group (especially with other presentations), independence of the elaboration and adherence to deadlines.
Editing tips
As a guideline for the preparation of a good seminar paper (incl. presentation and elaboration) the book
Markus Deininger and Horst Lichter and Jochen Ludewig and Kurt Schneider. Studien-Arbeiten: ein Leitfaden zur Vorbereitung, Durchführung und Betreuung von Studien-, Diplom- Abschluss- und Doktorarbeiten am Beispiel Informatik. 5th edition. vdf Hochschulverlag, Zurich, 2005.
which is available from the supervisor of the seminar (Prof. Dr. Andreas Hotho). We recommend the purchase of this book (9,50 €), because it can accompany you until the master thesis (and further). The grading of the seminar paper is done according to the scheme given there on page 77, adapted to the requirements of a seminar paper.