Universität Wien

052322 VU Graph Learning (2024W)

3.00 ECTS (2.00 SWS), SPL 5 - Informatik und Wirtschaftsinformatik
Prüfungsimmanente Lehrveranstaltung
Moodle
Di 26.11. 11:30-13:00 Seminarraum 6, Währinger Straße 29 1.OG

An/Abmeldung

Hinweis: Ihr Anmeldezeitpunkt innerhalb der Frist hat keine Auswirkungen auf die Platzvergabe (kein "first come, first served").

Details

max. 25 Teilnehmer*innen
Sprache: Englisch

Lehrende

Termine (iCal) - nächster Termin ist mit N markiert

  • Dienstag 01.10. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 08.10. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 15.10. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 22.10. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 29.10. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 05.11. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 12.11. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 19.11. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 03.12. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 10.12. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 17.12. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 07.01. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 14.01. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 21.01. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG
  • Dienstag 28.01. 11:30 - 13:00 Seminarraum 6, Währinger Straße 29 1.OG

Information

Ziele, Inhalte und Methode der Lehrveranstaltung

Machine learning and artificial intelligence are becoming increasingly important in industry and academia. Graph learning refers to methods suitable for data with a complex structure, such as social networks, molecules, knowledge graphs, or communication and transaction networks. It requires solving problems at the intersection of machine learning, graph theory, and algorithmics.

This course covers the fundamentals of graph learning as well as state-of-the-art methods and recent research directions, including graph neural networks (e.g., message-passing neural networks, spectral and recurrent methods, graph transformers), graph kernels (e.g., Weisfeiler-Leman kernels), and their relation.

Pen-and-paper exercises and programming assignments complement the lectures. Upon successful participation in the course, students will understand the fundamentals of graph learning, know how to apply basic methods in theory and practice, and will have the ability to connect to current research.

Art der Leistungskontrolle und erlaubte Hilfsmittel

* Written exam (individual work): at the end of the semester; you will be allowed to bring a handwritten A4 sheet (2 pages) of notes.

* Pen and paper exercises (individual work): you will solve pen and paper exercises at home; to be awarded credits for your solutions, you must present your solutions in the exercise sessions (you will be randomly selected).

* Programming assignments (individual or group work): you will solve graph learning programming assignments at home; you will have to submit your executable source code and a written report describing the results obtained with your implementation; you will have to present your results in in-person sessions.

* Paper presentation (individual or group work): You will choose a research paper on one of the course topics, understand it carefully, and present the key ideas to the course participants. The topic can be chosen from a list of selected papers published at the beginning of the course. Students are welcome to suggest research papers they wish to work on, but an instructor's prior agreement must be obtained.

Mindestanforderungen und Beurteilungsmaßstab

35% Written exam
25% Pen and paper exercises
20% Programming assignments
20% Paper presentation

P = Average weighted percentage on the written exam, the pen and paper exercises, the programming assignments, and the paper presentation

87% <= P <= 100% Sehr Gut (1)
75% <= P < 87% Gut (2)
63% <= P < 75% Befriedigend (3)
50% <= P < 63% Genügend (4)
0% <= P < 50% Nicht Genügend (5)

To successfully complete the course, you need to achieve at least 50% of the points on the written exam. Attendance of the lecture parts of the course is voluntary but highly recommended. Attendance of the pen and paper exercise sessions, programming assignment presentation, paper presentation, and the written exam is compulsory to obtain points.

Prüfungsstoff

All topics covered in class, the reading material, and the exercises. Referenced literature (as indicated in detail on the lecture slides).

Literatur

William L. Hamilton: Graph Representation Learning. Synthesis Lectures on Artificial Intelligence and Machine Learning, Morgan & Claypool Publishers 2020. https://doi.org/10.2200/S01045ED1V01Y202009AIM046

Michael M. Bronstein, Joan Bruna, Taco Cohen, Petar Veličković: Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges. https://arxiv.org/abs/2104.13478

Nils M. Kriege, Fredrik D. Johansson, Christopher Morris: A survey on graph kernels. Appl. Netw. Sci. 5(1): 6 (2020). https://arxiv.org/abs/1903.11835

Further material and recent research papers will be made available in Moodle.

Zuordnung im Vorlesungsverzeichnis

Letzte Änderung: Do 19.09.2024 10:25