Contents |
Course Code
IAM 557 (9700557)
Credit
(3-0) 3
Prerequisites
All students of the institute are welcome to this course.
Content/Aim
The objective of this course is to provide students with theory, methods and practice in data mining, inference, prediction and information. During and after the course, a deepening and continuation is offered by projects, e.g., in computational biology and medicine informatics), in financial mathematics (loan banking, risk management), in data/information processing and technology.
Various methods from statistics, discrete mathematics, numerical anlysis and information theory are presented and combined from the view-point of modern algorithms and applications. The computational aspect is taken into account. Throughout the course, we discuss and perform the practical means of simulation and learning. The purpose of the exercises is to familiarize the students with the most usual algorithmical and numerical techniques and their applications.
Learning Outcomes
At the end of the course students should have a good overview of modern methods in statistical learning and information. They should also be able to choose and, by calculation and simulation, work them out appropriately in contexts of applications.
Suggested Books
- N. Christianini and J. Shawe-Taylor, “An Introduction to Support Vector Machines”, Cambridge University Press, 2000.
- Th. M. Cover and J.A. Thomas, “Elements of Information Theory”, Wiley Series in Communication, 1991.
- T. Hastie, R. Tibshirani and J. Friedman, “The Elemenents of Statistical Learning”, Springer Series in Statistics, 2001.
- M. Laetsch, “Distance between Strings and Its Application to Amino Acid Sequences – An Information Theoretic Approach”, diploma thesis, Chemnitz University of Technology, Department of Mathematics.
- T.G. Oberstein, “Efficient Training of Observable Operator Models using Context Graphs”, diploma thesis, University of Cologne, Institute of Mathematics, GMD Report, 2001.
During the course, lecture notes (a manuscript) will be distributed. Furthermore, in lectures and exercises there will be further valuable appendices (e.g., elements of probability and statistics) and modern texts offered for updating basic knowledge and for treating interesting mini-projects, respectively.
Resources
- MATLAB 6.1
Outline
- Introduction into statistical learning and simulation
- Introduction into supervised learning
- Elements of MATLAB
- Linear methods of regression
- Linear methods of regression
- Linear methods in classification
- Linear methods in classification
- Model assessment and selection
- Model assessment and selection
- Model inference and averaging
- Model inference and averaging
- Additive models, trees and related methods
- Additive models, trees and related methods
- Prototype methods and nearest neighbours
- Prototype methods and nearest neighbours
- Cluster algorithms and support vector machines
- Unsupervised learning, and an outlook
Schedule
First Meeting in Fall 2008-2009
Tuesday, September 16, 9.40-12.30
in IAM S209