## Expected prior knowledge

Linear algebra, multivariate statistics.

## Course Contents

Recapitulation of multi-dimensional statistics, data visualisation, density esimation, cluster analysis.

Representation of real world objects by features, prototypes and dissimilarities.

Training pattern classifiers by examples.

Feature extraction.

Bayes' rule.

Classification by statistical discriminants, neural networks, decision trees or support vector machines.

Statistical learning theory.

One-class classifiers.

Combined appraoches.

EM algorithm.

Partially supervised learning.

Evaluation procedures, cross validation.

Overtraining, regularisation.

## Study Goals

After succesfully completing this course, the student is able to:

recognise pattern recognition problems and select algorithms to solve them;

read and comprehend recent articles in engineering-oriented pattern recognition journals, such as IEEE Tr. on PAMI;

construct a learning system to solve a given simple pattern recognition problem, using existing software.

## Timetable

See for the course, exam and resit schedule, the TU Delft timetable page.

## Education Method

Lectures, lab work

Workload is around 30 hours for attending lectures, 40 hours of reading study material and preparing lectures, 60 hours for the lab course, 20 hours for preparing the exam, 3 hours for the exam, and 8 hours for a final report (161 hours in total).

## Literature and Study Materials

S.Theodoridis and K.Koutroumbas, Pattern Recognition (2nd ed.), Elsevier, 2009, ISBN-978-1-59749-272-0;

Sheets;

PRTools user manual;

Pattern Recognition exercises with PRTools.

## Assessment

Homework, Computer laboratory assignment and written examination.

The final grade = 20% homework grade + 40% final computer lab. assignment + 40% written exam.

The exam is an open book examination, except for a multiple choice questions. Only the book can be used, no additional printout of slides, notes, etc. The exam takes 3 hours. You can use a graphical calculated during the exam. Further, no phones, no tables, no laptops or other electronic equipment.

For the homework exercises and the final computer lab there is no resit. For the exam there is a resit.

## Enrolment

You are required to subscribe in Brightspace, and be present at the first lecture.

## Remark

See also the course description in the TU Delft study guide 2018-2019.