Theoretical Concepts of Machine Learning (2VO)
| Course no.: | 365.041 |
| Lecturer: | Ulrich Bodenhofer |
| Start: | Tue, Oct 16, 2007 |
| Times: | Tue, Oct 16, 2007, 1:45-3:15pm Tue, Oct 23, 2007, 1:45-4:15pm Tue, Oct 30, 2007, 1:45-3:15pm Tue, Nov 6, 2007, 1:45-4:15pm Tue, Nov 13, 2007, 1:45-3:15pm Tue, Nov 20, 2007, 1:45-4:15pm Tue, Nov 27, 2007, 1:45-3:15pm Tue, Dec 4, 2007, 1:45-4:15pm Tue, Dec 11, 2007, 1:45-3:15pm Tue, Jan 8, 2007, 1:45-4:15pm Tue, Jan 15, 2007, 1:45-4:15pm Tue, Jan 22, 2007, 1:45-4:15pm Tue, Jan 29, 2007, 1:45-4:15pm |
| Location: | Kopfgebäude, room KG 712 |
| Mode: | VO, 2-3h, weekly |
| Registration: | KUSSS (after 15 Oct 2007, contact the lecturer) |
Motivation
Machine learning methods, i.e. methods that infer models/relationships by learning from data, are still gaining importance in various fields, such as, process modeling, speech and image processing, bioinformatics, and so forth. Their ability to cope with tasks for which no analytical model is available, ideally complements classical approaches. One has to acknowledge, however, that machine learning methods also bear great risks if they are applied inappropriately. The given lecture provides a look behind the curtain of machine learning. The goal is to make students acquainted with the basic concepts and methods to analyze, evaluate and understand models created by machine learning. In the sequel, we will also have a closer look at support vector machines and neural networks from this foundational perspective.Contents
- Repetition of the basic concepts of machine learning
- Evaluation criteria and optimization strategies
- Statistical learning theory
- Support vector machines: advanced topics and applications
- Neural networks: advanced topics and applications
- Selection of further topics
Necessary Background
Parts of the lecture will be quite mathematical, so a profound background in calculus, probability and statistics is necessary. This should not be a problem for graduate students of mathematics, computer science, physics, mechatronics, and statistics. Prior knowledge of machine learning (e.g. attendance of Prof. Widmer's lecture "Pattern Recognition and Classification") is surely helpful, but not an absolute pre-requisite. Due to the significant overlap with the lecture "Bioinformatics II: Theoretical Bioinformatics and Machine Learning", this lecture does not make sense for master students of bioinformatics.Course Material
Slides
- Introduction
[PDF] (Pages i-xii; 184KB; last update 2007-12-18) - Unit 1: Overview of Machine Learning
[PDF] (Pages 1-22; 225KB; last update 2007-10-17) - Unit 2: Model Evaluation in Supervised Machine Learning
[PDF] (Pages 23-105; 1.08MB; last update 2007-11-14) - Unit 3: Statistical Learning Theory
[PDF] (Pages 106-156; 450KB; last update 2007-12-11) - Unit 4: Support Vector Machines
[PDF] (Pages 157-291; 6.12MB; last update 2007-12-18) - Unit 5: Artificial Neural Networks
[PDF] (Pages 292-318; 618KB; last update 2008-01-22)
© 2007/2008 Ulrich Bodenhofer
This material, no matter whether in printed or electronic form, may be used for personal and educational use
only. Any reproduction of this material, no matter whether as a whole or in parts, no matter whether in printed or in
electronic form, requires explicit prior acceptance of the author.
Software demos
Notes for further reading
- Lecture Notes Bioinformatics II (PDF, 8.2MB)
Books recommended for further reading
- C. M. Bishop. Neural Networks for Pattern Recognition. Oxford University Press, 1995. ISBN 0-19-853864-2. [link]
- R. O. Duda, P. E. Hart, and D. G. Stork. Pattern Classification. Second edition. John Wiley & Sons, 2001. ISBN 0-471-05669-3. [link]
- T. Hastie, R. Tibshirani, and J. Friedman. The Elements of Statistical Learning. Springer, 2001. ISBN 0-387-95284-5. [link]
- R. Herbrich. Learning Kernel Classifiers. MIT Press, 2002. ISBN 0-262-08306-X. [link]
- B. Schölkopf and A. J. Smola. Learning With Kernels. MIT Press, 2002. ISBN 0-262-19475-9. [link]
- V. N. Vapnik. The Nature of Statistical Learning Theory. Springer, 1995. ISBN 0-387-98780-0. [link]
- V. N. Vapnik. Statistical Learning Theory. John Wiley &Sons, 1998. ISBN 0-471-03003-1. [link]