Learning Generative Models

Winter
(in German: Learning Generative Models
)
Module-ID: FIN-INF-120522
Abbreviation
LGM
Credit Points
6
Semester
Winter
Term
starting at 1.
Duration
1 Semester
Language
english
Level
Master
Link:
 LSF
Limited accessability:
Responsibility:
Sebastian Stober
Lecturer:
Sebastian Stober
Classes:
Lecture Learning Generative Models
Exercise class Learning Generative Models
 
Applicability in curriculum:
- M.Sc. INF: Informatik
- M.Sc. INGINF: Informatik
- M.Sc. WIF: Informatik
- M.Sc. DKE: Learning Methods and Models for Data Science
- M.Sc. DE: Methoden der Informatik
- M.Sc. VC: Computer Science
Intended learning outcomes:
Students can ..
confidently apply generative models to develop a solution for a given problem
follow recent publications on generative models and critically assess their contributions
formulate hypotheses and design & conduct experiments with generative models to validate them
document progress & design decisions for reproducibility and transparency
Content:
Training methods & architectures for generative models, in particular energy-based models (e.g. Restricted Boltzmann Machines, RBMs)
autoregressive models
variational learning and Generative Adversarial Nets (GANs)
flow-based models
diffusion models
Workload:
56 contact hours + 124h self study
Pre-examination requirements:
Type of examination:
Teaching method / lecture hours per week (SWS):
Oral exam (necessary preliminary work for admission to the examination is announced in the first week of the course)
Schein (necessary preliminary work for admission to the examination is announced in the first week of the course)
2 SWS lecture
2 SWS exercise class
Prerequisites according to examination regulations:
Recommended prerequisites:
keine
Introduction to Deep Learning or Deep Learning für Ingenieure (required)
Media:
Literature:
Ian Goodfellow, Yoshua Bengio & Aaron Courville: “Deep Learning”, MIT Press, 2016.
Comments: