This programming class complements courses on machine learning given in the school. It offers students the possibility to  understand some machine learning algorithms in depth by programming them and testing them rigorously. Students will be offered a choice of methods to program. Programming can be done in matlab or C/C++. Proper evaluation of machine learning will be stressed out. Students will learn about various methods to evaluate machine learning methods (crossvalidation, grid search, F-measure, ROC curve, etc) and will be asked to put these in practice.

Programming in matlab. Machine Learning. Statistics.

Students must have taken a machine learning course or follow one during the same semester. This programming class is meant to complement the Applied Machine Learning course, but can also complement other machine learning courses given at EPFL.

Students have at their disposal videos presenting the theory of the pendant course Applied Machine Learning as supplementary material.

Applied Machine Learning - MICRO-455

Pattern Classification and Machine Learning: CS-433

Data Analysis and Model Classification - EE-516

Basic notions in Machine Learning:

Supervised versus unsupervised learning

Classification, non-linear regression, clustering

• Produce code for steps of ML algorithms
• Develop a reasoning process to transform an algorithm into programming code

• Use both general and domain specific IT resources and tools

Computer-based practice session. Some short ex-cathedra lectures will be given at the beginning of the class.

Attendance to all sessions is necessary to progress rapidly and benefit from assistants' support.

The students are evaluated on pieces of code handed out regularly throughout the course.

• https://go.epfl.ch/MICRO-401https://go.epfl.ch/MICRO-401

• https://tube.switch.ch/channels/7e86d16dhttps://tube.switch.ch/channels/7e86d16d

Students must know how to program in Matlab (or C which is close and then be ready to learn Matlab).