Unit information: Pattern Analysis and Statistical Learning in 2012/13

Unit name	Pattern Analysis and Statistical Learning
Unit code	EMATM1400
Credit points	10
Level of study	M/7
Teaching block(s)	Teaching Block 2 (weeks 13 - 24)
Unit director	Dr. De Bie
Open unit status	Not open
Pre-requisites	EMAT10100 Engineering Mathematics 1, EMAT20200 Engineering Mathematics 2 (applied statistics), and EMAT20540 Discrete Mathematics 2 (or equivalents)
Co-requisites	None
School/department	Department of Engineering Mathematics
Faculty	Faculty of Engineering

Description including Unit Aims

Description: This unit provides first-hand experience about the problem of analysing complex real world datasets, like those provided by biology, web, engineering, and many other domains. Students will be exposed to the most recent approaches based on statistical methods, and optimization theory, and to state of the art algorithms. They will also experience real examples of data analysis, based on actual case studies.

Throughout this unit, the underlying principles behind pattern analysis algorithms and the statistical assessment of patterns will be emphasized.

Aims: To give students a broad understanding of concepts in pattern analysis and statistics as applied across a range of application domains. To give students first-hand experience in specific algorithms from statistical learning and pattern recognition, including kernel methods, probabilistic graphical models, string analysis, and more. To teach student the practical application of matlab to pattern analysis problems.

Intended Learning Outcomes

1. Students will access this unit with a basic knowledge of probability and will acquire working knowledge of practical data analysis, in real world situations.
2. They will be able to start from a set of data and deliver patterns and other relevant relations detected in it and assessments about their statistical significance.
3. They will learn general concepts about pattern analysis, that are valid in many domains: algorithmic and statistical principles to be used in different domains.
4. They will also acquire first hand experience in specific algorithms from statistical learning and pattern recognition, including kernel methods, probabilistic graphical models, string analysis, and more.
5. They will see how a real world data analysis task is performed, by practicing with real data and matlab.

Teaching Information

Lectures

Assessment Information

• 2-hour written examination 50% (learning outcomes 1-4)

• Coursework 50% (learning outcomes 1-5)

Reading and References

• Pattern Classification

Duda, Hart, Stork, Wiley, 2000

• The Elements of Statistical Learning

Hastie, Tibshirani, Friedman, Springer, 2001

• Kernel Methods for Pattern Analysis

Shawe-Taylor, Cristianini, Cambridge University Press, 2004.