Machine Learning in Signal Processing (MLSIP)5 ECTS
(englische Bezeichnung: Machine Learning in Signal Processing)

Lehrveranstaltungen:

• • Machine Learning in Signal Processing
    (Vorlesung, 3 SWS, Lehrbeauftragte et al., Blockveranstaltung 7.8.2017-11.8.2017 Mo-Fr, 8:15 - 11:30, H15; Block course in August 2017)
  • Supplements for Machine Learning in Signal Processing
    (Übung, 1 SWS, Christian Hümmer, Blockveranstaltung 2.10.2017-6.10.2017 Mo-Fr, 8:45 - 12:00, Raum n.V.; The room will be announced in due time.)

Empfohlene Voraussetzungen:

Recommended: Digital Signal Processing

Inhalt:

The goal of this lecture is to familiarize the students with the overall pipeline of a pattern recognition system. The various steps involved from data capture to pattern classification are presented. The lectures start with a short introduction, where the nomenclature is defined. Commonly used preprocessing methods are then described. A key component of pattern recognition is feature extraction. Thus, several techniques for feature computation will be presented including Walsh transform, Haar transform, linear predictive coding (LPC), wavelets, moments, principal component analysis (PCA) and linear discriminant analysis (LDA). The lectures conclude with a basic introduction to classification. The principles of statistical, distribution-free and non-parametric classification approaches will be presented. Within this context we will cover Bayesian and Gaussian classifiers, as well as artificial neural networks.

Lernziele und Kompetenzen:

The students

• explain the general pipeline of a pattern recognition system

• apply various vector quantization methods

• apply histogram equalization and histogram stretching

• compare different thresholding methods

• apply the principle of maximum likelihood estimation to Gaussian probability density functions

• apply various low- and high-pass filters, as well as non-linear filters (homomorphic transformations, cepstrum, morphological operations, rank operations)

• apply various normalization methods

• understand the curse of dimensionality

• explain various heuristic feature extraction methods, e.g.projection to orthogonal bases (Fourier transform, Walsh/Hadamard transform, Haar transform), Linear Predictive Coding, geometric moments, feature extraction via filtering, wavelets)

• understand analytic feature extraction methods, e.g. Principal Component Analysis, Linear Discriminant Analysis

• define the decision boundary between classes

• compare different objective functions for feature selection

• explain the principles of statistical classification (optimal classifier, cost functions, Bayes classifier)

• understand different classifiers (Gauss classifier, polynomial classifier, non-parametric classifiers such as k-nearest neighbor classifier, Parzen windows, neural networks) and compare them w.r.t. their decision boundaries, their computational complexity, etc.

Literatur:

• lecture slides

• Heinrich Niemann: Klassifikation von Mustern, 2. überarbeitete Auflage, 2003

• Sergios Theodoridis, Konstantinos Koutroumbas: Pattern Recognition, 4th edition, Academic Press, Burlington, 2009

• Richard O. Duda, Peter E. Hart, David G. Stock: Pattern Classification, 2nd edition, John Wiley & Sons, New York, 2001

Weitere Informationen:

Verwendbarkeit des Moduls / Einpassung in den Musterstudienplan:
Das Modul ist im Kontext der folgenden Studienfächer/Vertiefungsrichtungen verwendbar:

1. Advanced Signal Processing & Communications Engineering (Master of Science)
   (Po-Vers. 2016w | TechFak | Advanced Signal Processing and Communications Engineering (Master of Science) | Masterprüfung | Pflichtmodule | Machine Learning in Signal Processing)
2. Communications and Multimedia Engineering (Master of Science)
   (Po-Vers. 2011 | TechFak | Communications and Multimedia Engineering (Master of Science) | Masterprüfung | Wahlpflichtmodule | Technische Wahlpflichtmodule)
3. Communications and Multimedia Engineering (Master of Science)
   (Po-Vers. 2011 | TechFak | Communications and Multimedia Engineering (Master of Science) | Masterprüfung | Wahlmodule | Technische Wahlmodule)
4. Informations- und Kommunikationstechnik (Master of Science)
   (Po-Vers. 2016s | TechFak | Informations- und Kommunikationstechnik (Master of Science) | Schwerpunkte im Masterstudium | Schwerpunkt Eingebettete Systeme | Wahlpflichtmodule | Wahlpflichtmodul aus EEI im Schwerpunkt Eingebettete Systeme)
5. Informations- und Kommunikationstechnik (Master of Science)
   (Po-Vers. 2016s | TechFak | Informations- und Kommunikationstechnik (Master of Science) | Schwerpunkte im Masterstudium | Schwerpunkt Kommunikationsnetze und Übertragungstechnik | Wahlpflichtmodule | Wahlpflichtmodul aus EEI im Schwerpunkt Kommunikationsnetze)
6. Informations- und Kommunikationstechnik (Master of Science)
   (Po-Vers. 2016s | TechFak | Informations- und Kommunikationstechnik (Master of Science) | Schwerpunkte im Masterstudium | Schwerpunkt Multimediasysteme | Wahlpflichtmodule | Wahlpflichtmodul aus EEI im Schwerpunkt Multimediasysteme)

Studien-/Prüfungsleistungen:

Machine Learning in Signal Processing (Prüfungsnummer: 84401)

Prüfungsleistung, Klausur, Dauer (in Minuten): 90, benotet, 5 ECTS

Anteil an der Berechnung der Modulnote: 100.0 %

Prüfungssprache: Englisch

Erstablegung: SS 2017, 1. Wdh.: WS 2017/2018

1. Prüfer:

Walter Kellermann

Termin: 12.10.2017, 14:00 Uhr, Ort: LS
Termin: 04.04.2018, 10:00 Uhr, Ort: LS