Structured Model Learning
- Department of Cybernetics
This advanced machine learning course covers learning and parameter estimation for structured models like Markov Random Fields, Belief Networks and (stochastic) Deep Neural Networks.
- Solid knowledge of of statistical machine learning (cf. BE4M33SSU)
- Basic knowledge of Graphical Models (cf. XEP33GMM)
- Syllabus of lectures:
(1) Markov Random Fields & Gibbs Random Fields
(2) Belief Networks & Stochastic Neural Networks
(3) Learning of structured output classifiers by Perceptron
(4) Structured Output Support Vector Machines
(5) Learning max-sum classifiers by SO-SVM
(6) Optimization methods for SO-SVM
(7) Maximum Likelihood learning for MRFs
(8) Variational Autoencoders
(9) Variational Bayesian inference for DNNs
(10) Generative adversarial networks
- Syllabus of tutorials:
The seminars will be dedicated to discussions and deepening the knowledge acquired at the lectures.
- Study Objective:
The course aims to communicate knowledge on theory and algorithms for the two currently most successful branches of structured model learning - statistical learning and structured output learning.
- Study materials:
1. B. Taskar, C. Guestrin, and D. Koller. Maximum-margin markov networks. In Advances in Neural Information Processing Systems. MIT Press, Cambridge, MA, 2004.
2. I. Tsochantaridis, T. Joachims, T. Hofmann, and Y. Altun. Large margin methods for structured and interdependent output variables. Journal of Machine Learning Research, 6:1453-1484, Sep. 2005.
3. V. Franc and B. Savchynskyy. Discriminative learning of max-sum classifiers. Journal of Machine LearningResearch, 9(1):67-104, January 2008. ISSN 1532-4435.
4. M.J. Wainwright and M.I. Jordan. Graphical Models, Exponential Families, and Variational Inference. Foundations and Trends in Machine Learning, 1(1-2):1-305, 2008.
- Further information:
- No time-table has been prepared for this course
- The course is a part of the following study plans: