Duration:
1 Semester | Turnus of offer:
irregularly | Credit points:
5 |
Course of studies, specific field and terms: - Master CLS 2023 (optional subject), mathematics, 2nd or 4th semester
- Bachelor CLS 2023 (optional subject), mathematics, 6th semester
- Master CLS 2016 (optional subject), mathematics, 2nd or 4th semester
- Bachelor CLS 2016 (optional subject), mathematics, 6th semester
|
Classes and lectures: - Non-smooth Optimization and Analysis (exercise, 1 SWS)
- Non-smooth Optimization and Analysis (lecture, 2 SWS)
| Workload: - 30 Hours work on project
- 65 Hours private studies and exercises
- 45 Hours in-classroom work
- 10 Hours exam preparation
| |
Contents of teaching: | - Introduction to non-smooth analysis: convexity, subdifferentials, existence, Legendre- Fenchel conjugate, duality
- First- and higher-order numerical optimization methods: PDHG and interior-point methods
- Approximation of discrete and non-convex problems
- Generalized derivatives and Clarke subdifferential, semismooth Newton methods
- Applications in image processing and computer vision
| |
Qualification-goals/Competencies: - The students understand the strengths of non-smooth models.
- They can devise and analyse models for simple problems.
- They understand the advantages, disadvantages, and application areas of each optimization method.
- They know how to select and specialize a suitable optimization method for a given model.
- Interdisciplinary qualifications:
- Students have advanced skills in modeling.
- They can translate theoretical concepts into practical solutions.
- They are experienced in implementation.
- They can think abstractly about practical problems.
|
Grading through: - Written or oral exam as announced by the examiner
|
Requires: |
Responsible for this module: Teachers: |
Literature: - Rockafellar, Wets: Variational Analysis - Springer
- Boyd, Vandenberghe: Convex Optimization - Cambridge University Press
- Ben-Tal, Nemirovski: Lectures on Modern Convex Optimization - SIAM
- Paragios, Chen, Faugeras: Handbook of Mathematical Models in Computer Vision - Springer
|
Language: - German and English skills required
|
Notes:Prerequisites for attending the module: - None (Familiarity with the topics of the required modules is assumed, but the modules are not a formal prerequisite for attending the course). Prerequisites for the exam: - Homework assignments and their presentation are ungraded examination prerequisites which have to be completed and positively evaluated before the first examination. Examination: - MA5035-L1: Non-smooth Optimization and Analysis, written examination (90min) or oral examination (30 min) as decided by examiner, 100 % of final mark |
Letzte Änderung: 22.2.2022 |
für die Ukraine