Module Description

Module: Pattern Recognition and Data Compression

Courses:

TitleTypeHrs/WeekPeriod
Pattern Recognition and Data CompressionLecture4Summer Semester

Module Responsibility:

Prof. Rolf-Rainer Grigat

Admission Requirements:

None

Recommended Previous Knowledge:

Linear algebra (including PCA, unitary transforms), stochastics and statistics, binary arithmetics

Educational Objectives:

Professional Competence

Theoretical Knowledge

Students can name the basic concepts of pattern recognition and data compression.

Students are able to discuss logical connections between the concepts covered in the course and to explain them by means of examples.

Capabilities

Students can apply statistical methods to classification problems in pattern recognition and to prediction in data compression. On a sound theoretical and methodical basis they can analyze characteristic value assignments and classifications and describe data compression and video signal coding. They are able to use highly sophisticated methods and processes of the subject area. Students are capable of assessing different solution approaches in multidimensional decision-making areas.

Personal Competence

Social Competence

k.A.

Autonomy

Students are capable of identifying problems independently and of solving them scientifically, using the methods they have learnt.

ECTS-Credit Points Module:

6 ECTS

Examination:

Written exam

Workload in Hours:

Independent Study Time: 124, Study Time in Lecture: 56


Course: Pattern Recognition and Data Compression (Lecture)

Lecturer:

Rolf-Rainer Grigat

Language:

English

Period:

Summer Semester

Content:

Structure of a pattern recognition system, statistical decision theory, classification based on statistical models, polynomial regression, dimension reduction, multilayer perceptron regression, radial basis functions, support vector machines, unsupervised learning and clustering, algorithm-independent machine learning, mixture models and EM, adaptive basis function models and boosting, Markov random fields

Information, entropy, redundancy, mutual information, Markov processes, basic coding schemes (code length, run length coding, prefix-free codes), entropy coding (Huffman, arithmetic coding), dictionary coding (LZ77/Deflate/LZMA2, LZ78/LZW), prediction, DPCM, CALIC, quantization (scalar and vector quantization), transform coding, prediction, decorrelation (DPCM, DCT, hybrid DCT, JPEG, JPEG-LS), motion estimation, subband coding, wavelets, HEVC (H.265,MPEG-H)

Literature:

Schürmann: Pattern Classification, Wiley 1996
Murphy, Machine Learning, MIT Press, 2012
Barber, Bayesian Reasoning and Machine Learning, Cambridge, 2012
Duda, Hart, Stork: Pattern Classification, Wiley, 2001
Bishop: Pattern Recognition and Machine Learning, Springer 2006

Salomon, Data Compression, the Complete Reference, Springer, 2000
Sayood, Introduction to Data Compression, Morgan Kaufmann, 2006
Ohm, Multimedia Communication Technology, Springer, 2004
Solari, Digital video and audio compression, McGraw-Hill, 1997
Tekalp, Digital Video Processing, Prentice Hall, 1995