Decision-Theoretic Machine Learning

News

The aim and the scope of the course

The aim of the course:

To explain theoretical foundations of machine learning in order to show how simple algorithms can be used for solving complex problems

The scope of the course:

• Introduction to applications and theory of machine learning,
• An overview of the basic learning algorithms (lazy learning, decision trees, generative models, discriminative models, linear models),
• Learning theory (loss function, risk, regret, empirical risk minimization, generalization bounds, calibration),
• Learning to rank (performance measures, learning theory for ranking problems, reduction to binary classification),
• Multi-label classification (performance measures, learning theory for multi-label classification, learning reductions for multi-label classification).

Information about the course

Time and Place

• Lectures: April 25, May 9, 16 and 30, 15:10-18:20, lecture room 122 BT

Schedule of Lectures

Evaluation

In order to pass the course you need to solve some of the problems described in the pdf below from 4 different topics. For each solved problem you can get max. 1 point. However, you cannot get more than 1 point from a given topic. The final mark will be given according to the following rule:

• 3.0 points - 5.0
• 2.0 points - 4.0
• 1.0 points - 3.0

Your solutions should be sent (in a LaTeX-generated PDF file) to both instructors via email. Please use a tag ‘[DTML]’ in the title.

The deadline is June 30, 2019.

Description of problems: [pdf]

Bibliography

T. Hastie, R. Tibshirani, J. Friedman, Elements of Statistical Learning: Second Edition. Springer, 2009.
http://www-stat.stanford.edu/~tibs/ElemStatLearn/

Y. S. Abu-Mostafa, M. Magdon-Ismail, H-T. Lin, Learning From Data.
http://amlbook.com

D. Barber. Bayesian Reasoning and Machine Learning. Cambridge University Press, 2012.
http://www.cs.ucl.ac.uk/staff/d.barber/brml/

Christopher M. Bishop. Pattern Recognition and Machine Learning. Springer-Verlag, 2006.