Manifold Optimization for Representation Learning

What it is about

In this module you will learn how to extract representations from data via optimization on matrix manifolds. In the end you will be able to describe central models of Representation Learning and derive and implement simple optimization algorithms on manifolds.

What you will be able to do

• describe fundamental models of Representation Learning
• explain the constraints of the relevant manifolds
• understand technical concepts of optimization on manifolds
• derive simple optimization algorithms on manifolds
• implement classical algorithms for different Representation Learning approaches

What the module consists of

• VorlesungIntroduction to models and concepts of Representation Learning and manifold optimization
• Übungen/TutorialsDiscussion of tasks and programming exercises; support with solutions

Teaching method

• Frontalunterricht (Tafel, Beamer)Conveying definitions, theory and examples
• GruppendiskussionenDeveloping new definitions and concepts on simple examples
• Tutorials with exercises and programming tasksDeepening understanding and practical application