På Svenska

EN2202 Pattern Recognition 7.5 credits

Mönsterigenkänning

How can you make a computer understand your voice? How can you make a computer understand your handwriting? How do you detect signal patterns that are hidden in noise? How can a computer distinguish between ECG recordings from healthy and sick hearts? The course in Pattern Recognition gives you the theory to answer this kind of questions. In the course project, you create your own MatLab toolbox for pattern recognition.

  • Educational level

    Second cycle

  • Academic level (A-D)

  • Subject area

    Electrical Engineering

  • Grade scale

    A, B, C, D, E, FX, F

Course offerings

Learning outcomes

The participants shall after the course be able to

* design systems and algorithms for pattern recognition (signal classification), with focus on sequences of patterns that are analyzed using, e.g., hidden Markov models (HMM),

* analyse classification problems probabilistically and estimate classifier performance,

* understand and analyse methods for  automatic training of classification systems,

* apply Maximum-likelihood parameter estimation in relatively complex probabilistic models, such as mixture density models and hidden Markov models,

* understand the principles of Bayesian parameter estimation and apply them in relatively simple probabilistic models.

Course main content

The course is about the theoretical foundation of pattern recognition and gives an introduction to technical applications, especially in speech recognition and image or sound classification.

Disposition

Lectures (24h), tutorials (24h), and project homework.

Eligibility

For single course students: 120 credits and documented proficiency in English B or equivalent

Prerequisites

  • SF1901 Probability Theory and Statistics, or equivalent.
  • EQ1220 Signal Theory or equivalent is recommended but not required.

Literature

Arne Leijon (20xx) Pattern Recognition. KTH. (latest version)

Examination

  • INL1 - Home Work, 2.5 credits, grade scale: A, B, C, D, E, FX, F
  • TEN1 - Examination, 5.0 credits, grade scale: A, B, C, D, E, FX, F

Written exam and compulsory Homework Assignment including Matlab implementation of classifier tools.

Requirements for final grade

Exam 5p (grade A-F). Homework Assignment 2.5p (A-F). The final grade is a weighted sum of graded performance on the Exam and Homework Assignment, with weight 25 for the exam and 10 for the Homework Assignment.

Offered by

EES/Sound and Image Processing

Contact

Saikat Chatterjee <sach@kth.se>

Examiner

Markus Flierl <mflierl@kth.se>

Mikael Skoglund <skoglund@kth.se>

Supplementary information

All course material in English. Student assignment reports in either Swedish or English.

Version

Course plan valid from: Autumn 10.
Examination information valid from: Autumn 10.