| Instructors | Dr. Pavel Laskov, Dr. Blaine Nelson |
| Class meetings | Tue, 14:00 s.t. - 16:00 |
| Location | Sand, kleiner Hörsaal, F122 |
| Credit points, Diplom | 2 SWS (lectures) + 1 SWS exercizes |
| Credit points, Master | 4 LP (lectures + exercizes) |
| Office hours | by appointment |
| Instruction language | English |
| Examination area | Practical informatics |
The course offers an overview of advanced machine learning methods and applications. We will focus on so-called "kernel methods", in which the learning problems are formulated via paiwise relationships between data points. After an introduction of the underlying mathematical properties of kernel functions, various algorithms and practical applications of kernel methods will be considered. The course requires a basic knowledge of machine learning methods and the underlying mathematical and statistical instruments.
Diploma students can request an examination on this course in a usual manner and in any reasonable combination with another course. The amount of work covered by this course constitute 2 SWS for lectures and 1 SWS for graded exercises. An exercise certificate will be issued at the end of the semester. The grade for master students will be composed of the results of the written final exam (70%) and exercise grades (30%).
John Shawe-Taylor and Nello Cristianini: Kernel Methods for Pattern Analysis . Cambridge University Press, 2004.
| Date | Location | Topic | Instructor | Slides |
|---|---|---|---|---|
| Tue, 17.04 | F122 | Course Introduction, application examples | Nelson | Lecture 1 |
| Tue, 24.04 | F122 | Overview of kernel methods | Laskov | Lecture 2 |
| Tue, 08.05 | F122 | Kernel functions: mathematical foundations | Nelson | Lecture 3 |
| Tue, 15.05 | F122 | Elementary kernel algorithms | Laskov | Lecture 4 |
| Tue, 22.05 | F122 | Eigendecompositions, kernel PCA and CCA | Nelson | Lecture 5 |
| Tue, 05.06 | F122 | Empirical risk minimization, Support Vector Machines | Nelson | Lecture 6 |
| Tue, 12.06 | F122 | "Big" learning | Laskov | Lecture 7 |
| Tue, 19.06 | F122 | Online and incremental learning | Laskov | Lecture 8 |
| Tue, 26.06 | F122 | NO CLASS | ||
| Tue, 03.07 | F122 | Kernel methods for structured inputs | Laskov | Lecture 9 |
| Tue, 10.07 | F122 | Kernel methods for structured outputs | Laskov | Lecture 10 |
| Tue, 17.07 | F122 | Learning in adversarial environments | Nelson | Lecture 11 |
| Tue, 24.07 | F122 | Classifier evasion methods, adversarial games | Nelson | Lecture 12 |
| Tue, 31.07 | F122 | Final exam |
