Advanced Topics in Machine Learning

This web page provides information on the course Advanced Topics in Machine Learning (summer term 2019). The course deals with selected topics of Machine Learning, including:

  • Support Vector Machines
  • Semi-Supervised Learning (semi-supervised classification and clustering)
  • Dealing with massive datasets

Prerequisite for attending this course is a basic knowledge of computer science, especially in Machine Learning. Programming skills are an advantage concerning the practical exercises.

Course Schedule and Room Assignments

Title Time Start Room
Lecture Thursday 3:00pm - 5:00pm 04.04.2019 G22A-020
Exercise (Group 1) Monday 09:00am - 11:00am 08.04.2019 G22A-110
Exercise (Group 2) Wednesday 1:00pm - 3:00pm 10.04.2019 G22A-120
Exam Consultation Friday 1:00pm - 3:00pm
Wednesday 1:00pm - 3:00pm


Exam Thursday 12:00pm - 2:00pm 25.07.2019 H1
Exam Review

Please ask via mail for a time slot in the following weeks:

26th of August 2019 until the 6th of September 2019


14th of October 2019 until the 1st November 2019


Further information on the lecture and the exercise can be found in the LSF portal.

Course Staff

If you have any questions concerning the lectures or assignments please contact (preferably by email):

Exercise Classes

The exercise classes have two objectives. First, regular assignments concerning the theory taught in the lecture will be given (about one week in advance). These have to be prepared by the students and are then discussed during class. Secondly, the lecture will be accompanied by a software project. Its goal is to practice the implementation of machine learning techniques into a larger system. This will be done as a joint group work. The development will partly be done during the exercise classes. However, further development outside the class might be necessary to complete the project. We expect active involvement of all students, both in the project and the theoretical assignments.

Please register yourself via LSF for the exercise classes. (Will open on the 18th of March 2019)

Requirements for Class Fulfillment

At the end of the course, there will be an oral exam. As a prerequisite, we expect active involvement both during the exercise and in the software project.


We will provide lecture slides, assignment sheets, and further material during the course.

Lecture Slides
Exercise Material
Further Material
  • Project Assignment due by 13th of May (concept) and 19th of June (report + source code), presentations on 24th/26th of June and 1st/3rd of July
  • Doodle for presentation times!
  • Doodle for exam preparation dates!
  • EUR-Lex data set by TU Darmstadt
  • Haussler, D. (1988). Quantifying Inductive Bias: AI Learning Algorithms and Valiant's Learning Framework. Artif. Intell., 36, 177-221. Link
  • N. Cristianini and J. Shawe-Taylor: An Introduction to Support Vector Machines, Cambridge University Press, Cambridge, 2000. (especially the first four chapters)
  • A. J. Smola, P. Bartlett, B. Schölkopf, and D. Schuurmans: Advances in Large Margin Classifiers, MIT Press, Cambridge, MA, 2000.
  • B. Schölkopf and A. J. Smola: Learning with Kernels, MIT Press, Cambridge, MA, 2002.
  • J. Shawe-Taylor and N. Cristianini: Kernel Methods for Pattern Analysis, Cambridge University Press, Cambridge, 2004.

Frequently Asked Questions 

  • Question: Do I have to register for the course? If so, where?
    Answer: Registrations are done via LSF (link above). The registrations will be open from the 18th of March 2019 until the 26th of April.
  • Question: I can't be there for the first exercise! What should I do?
    Answer: Please write to Marcus Thiel or Sayantan Polley (and best both) an e-mail about your absence.
  • Question: Do I really have to do the first assignment for the FIRST exercise?
    Answer: Yes, of course. The questions are easy enough to be handled without having heard the lecture.
  • Question: One exercise seems hardly enough. Shouldn't there be more?
    Answer: Additional exercises will be planned based on the number of interested students. Most likely in or after the first lecture.
  • Question: What's that thing about "voting"?
    Answer: It means, that you have to prepare the assignments BEFORE the exercise, in which they are discussed in. A sheet will be given around, where you have to mark all assignments, that you were able to solve. Marking an assignment means, that you are able to present your solution in front of the class. I expect everyone to at least present once, but twice would be best. Your solution does not have to be correct, but if you are not able to explain the assignment at all, all marked tasks for the particular exercise will be removed from your votings.
  • Question: How many "votings" do I need in order to get my exam admission?
    Answer: The absolute minimum is half of all the assignments. Every task on each assignment sheet counts as a single point. So you have to solve half of these tasks. BUT I expect active involvement in more than half of the tasks. Please try to solve at least 2/3. Achieving less than half is not acceptable and an admission without more than half is NOT up to discussion.
  • Question: Is this all I need for my admission?
    Answer: No! There is still the project, which has to be completed until the end of the exercise classes. (a bit earlier than that) Exact dates will be given in the first exercises.
  • Question: What is that project?
    Answer: The actual project will be given in the first or second exercise. It will be about data analysis and will involve a certain amount of programming, The latter can be greatly reduced by using existing libraries and tools.
  • Question: But I'm not from a Computer Science background / I can't program! Can I do something else instead?
    Answer: No, the project will be equal for everyone. The way you solve it can differ though.
  • Question: I have never heard anything about Data Mining and/or Machine Learning before. Should I still visit this course?
    Answer: Only if you are willing to fill the gaps yourself. This course requires a certain basic understanding about Machine Learning techniques. You don't need to be an expert though. You can have a look at the literature at the Machine Learning course.
  • Question: My math skills are very rusty, do I need them for this course?
    Answer: Same as the Machine Learning question: You need to have basic skills (e.g. derivations, matrices, vectors) in order to understand this course. Those things will only partially be explained again, so you should be ready to fill the gaps yourself.
  • Question: I really, really like Machine Learning, but the assignments are way too difficult and time consuming! Could we please reduce the amount of work?
    Answer: At the moment: No. The assignments may change depending on the feedback though.
  • Question: How will the exam look like? Will it be oral or written?
    Answer: The exam will be written. An exam preparation date will be scheduled close to the end of the lecture time.

Last Modification: 23.03.2020 - Contact Person: