Technische Universität München Robotics and Embedded Systems
 

Praktikum Maschinelles Lernen

 
Veranstalter Dipl.-Inf. Christian Osendorfer
Modul IN8903
Typ Praktikum
Semester SS 2008
ECTS 10.0
SWS 6
Vorbesprechung -
Zeit & Ort Di 16:00 - 18:00 MI 03.07.023
Schein erfolgreiche Teilnahme am Praktikum

About

The lab course is over!

We will start with ancient statistical techniques such as Bayes classifiers and Linear Discriminant Analysis (LDA) as well as more recently established methods such as feed forward Neural Networks and Hidden Markov Models.

The area of Machine Learning has grown tremendously over the past 15 years, and lots of new approaches have been developed in this period. Some of these (Support Vector Machines, Long Short Term Memory, Independent Component Analysis) will also be treated in this course.

There will be about 10 assignments during the semester. Each will be discussed in a one hour meeting taking place once every week. You are supposed to solve the assignments in groups of 2 or 3 people. Each assignment is centered around the understanding and implementation of one specific machine learning technique.

In order to test your implementations, the assignments will come with data sets from meaningful applications.

Note

The course belongs to Theoretische Informatik and Technische Informatik (this is important only if you are enrolled in the Diplomstudiengang Informatik).

Prerequisites

You should be familiar with the contents of Analysis I/II, Linear Algebra I/II and Probability Theory. See also the next remark!

Programming

This course is mainly about implementing machine learning algorithms. It is not about learning how to program. You are supposed to know the basics of programming. You are free to choose your programming language and environment, i.e. you could use also MatLab or Octave.

Literature

There are lots of books on Machine Learning. Yet every assignment is self-contained; therefore books should not be necessary (but might, of course, be valuable addenda). A recommended classic: