Course OML: Optimization and Machine Learning

Dates and time:

Mondays 13.15 - 15.00 (March 6 - April 3 and April 17 - May 15)

Location:

All LNMB courses can be attended on the Campus Utrecht Science Park. The lecture room for the first week is different from all other weeks:

March 6: Buys Ballot building - lecture room 161
March 13, 20, 27; April 3, 17, 24; May 1, 8, 15: Minnaert building - lecture room 2.02

Note that there's no lecture on April 10 (Easter)

Coordinating lecturer :

Prof.dr. I. Birbil (University of Amsterdam)

Course description:
This course is both about the important role of optimization in Machine Learning, and on the role of Machine Learning to improve optimization methods.

The first five weeks (March 6 - April 3) will be taught by Prof.dr. Ilker Birbil (University of Amsterdam). He will give an introduction on supervised learning, with a special focus on the role of optimization:

Linear Models and Regularization: linear-ridge-logistic regression, Lasso, elastic net, logistic regression
Support Vector Machines: primal and dual models, kernel trick
Neural Networks: backpropagation, stochastic gradient descent and its variants
Trees and Forests: optimal classification trees, subset selection, rule generation
Boosting: margin maximization, gradient boosting, relation to duality.

The remaining four weeks are on specific research projects on Optimization and Machine Learning, and they use the techniques introduced in the first part of the course (more details will be added before the course starts):

April 17: Online Optimization under Predictions (Prof.dr. Leen Stougie, CWI)
April 24: Offline Optimization with Predictions (Prof.dr. Guido Schäfer, CWI)
May 1: Worst-case Analysis of the (Stochastic) Gradient Method (Prof.dr. Etienne de Klerk, Tilburg University)
May 8: Optimization with Constraint Learning (Prof.dr.ir. Dick den Hertog, University of Amsterdam)

The lecture on May 15 serves as a backup, should one of the lectures need to be postponed unexpectedly.

Prerequisites:
Integer and linear optimization, and basic knowledge on nonlinear optimization.

Literature:
Handouts.

Examination:
Take home problems.

Address of the coordinating lecturer:
Prof.dr. S.I. Birbil
Faculty of Economics and Business, Section Business Analytics
University of Amsterdam
E-mail: s.i.birbil@uva.nl