Universität Wien

053612 VU Optimisation Methods for Data Science (2024W)

4.00 ECTS (3.00 SWS), SPL 5 - Informatik und Wirtschaftsinformatik
Continuous assessment of course work
Moodle

Registration/Deregistration

Note: The time of your registration within the registration period has no effect on the allocation of places (no first come, first served).

Details

max. 25 participants
Language: English

Lecturers

Classes (iCal) - next class is marked with N

  • Wednesday 02.10. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 07.10. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 09.10. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 14.10. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 16.10. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 21.10. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 23.10. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 28.10. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 30.10. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 04.11. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 06.11. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 11.11. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 13.11. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 18.11. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 20.11. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 25.11. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 27.11. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 02.12. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 04.12. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 09.12. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 11.12. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 16.12. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 08.01. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 13.01. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 15.01. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 20.01. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 22.01. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02
  • Monday 27.01. 09:00 - 11:15 Hörsaal 3 Oskar-Morgenstern-Platz 1 Erdgeschoß
  • Monday 27.01. 09:45 - 11:15 PC-Seminarraum 1, Kolingasse 14-16, OG01
  • Wednesday 29.01. 09:45 - 11:15 PC-Seminarraum 3, Kolingasse 14-16, OG02

Information

Aims, contents and method of the course

The lectures will be held in presence.

This course will give an overview of modern optimization methods and their applications in data science.

Contents:
- Fundamentals of convex analysis
- First-order methods: gradient descent, subgradient method, acceleration, adaptivity, etc.
- Stochastic first-order methods: stochastic gradient descent, variance reduction.
- Higher-order methods: Newton's method, quasi-Newton method.

Assessment and permitted materials

(1) a written exam at the end of the semester (in person).
(2) 2-3 long homeworks during semester.
(3) 2 quizzes (about 15 minutes each) during the semester (announced at least one week in advance).
(4) bonus points for active participation during classes.

The exam will take place at the end of January, the exact date will be announced later.

Minimum requirements and assessment criteria

Exam: 50%
Homeworks: 40%
Quizzes: 10%

Precentage/Grades:

0-53: nicht genuegend/fail (5)
54-65: genuegend/pass (4)
66-77: befriedigend/satisfactory (3)
78-89: gut/good (2)
90-100: sehr gut/excellent (1)

Examination topics

all material covered during lectures

Reading list

1. A. Beck "First-Order Methods in Optimization".
2. Optimization for Machine Learning lecture notes by Martin Jaggi EPFL and Bernd Gärtner, ETH
https://raw.githubusercontent.com/epfml/OptML_course/master/lecture_notes/lecture-notes.pdf
3. Stephen Boyd and Lieven Vandenberghe. "Convex Optimization", https://web.stanford.edu/~boyd/cvxbook/.

Association in the course directory

Modul: OMD

Last modified: Tu 19.11.2024 09:45