Course title Selected Advanced Topics in Machine Learning
Assesment method Exam Hours per semester 60 Lect. Exercises Lab. Project
ETCS 4 Hours/week 2 2
Prerequisites
Thorough knowledge of basic Machine Learning and statistical techniques.
Course description
The course will cover selected advanced topics in Machine Learning which are currently active research areas. Topics will include: graphical models (Bayesian and Markov networks, Gibbs sampling), Bayesian statistics, conjugate priors, Latent Dirichlet Allocation, Hidden Markov Models, particle filters, learning with structured data, multi-instance and transfer learning, deep learning.
Course objectives
After completing the course, the students will be familiar with selected advanced topics in Machine Learning. The course will also aid the students in their PhD. Research related to Machine Learning.
Grading
The final grade will be given based on homeworks and a final exam.
Reference Texts and Software
Literature:
  1. MacKay D., Information Theory, Inference, and Learning Algorithms, Cambridge University Press, 2003.
  2. Additionally the course will use selected important publications from Machine Learning journals and conferences.
Software:
  1. R: http://www.r-project.org
  2. JAGS: http://mcmc-jags.sourceforge.net
Lecture Schedule
1. Graphical models. Bayesian networks. D-separation.
2. Exact inference in Bayesian networks.
3. Approximate inference in Bayesian networks.
4. Gibbs sampling.
5. Graphical Models and Machine Learning. JAGS.
6. Bayesian statistics. Conjugate priors.
7. Gibbs sampling improvements and implementation details.
8. Text Modeling. Latent Dirichlet Allocation.
9. Markov Networks.
10. Dynamic Bayesian Networks. Hidden Markov Models.
11. Bayesian filtering. Particle filters.
12. Structured Learning.
13. Transfer and multitask learning.
14. Deep learning.