|
04901 Advanced Digital Signal Processing
| |
Danish title: Advanced Digital Signal Processing
| |
Language: English
Credit points: 2,85 |
|
| Type: | Ph.D.-level, Open University
Language: English |
| |
| |
| |
| |
| |
| |
Recommended semester: 7th - 9th semester
| |
Scope and form: Lectures, exercises (Matlab), short presentations by participants.
| |
Examination: Evaluation of report(s) (13-scale)
| |
| |
| |
| |
| |
Department: Informatics and Mathematical Modelling
|
Aim: Design of neural networks:
Lars Kai Hansen. The objective is to provide the participants with operational experience in neural net training and design of networks for simple pattern recognition tasks. Keywords are; feed-forward networks, learning algorithms, network pruning and cross-validation.
Signal processing with neural networks: Jan Larsen.
The objective is provide the participants with methods for design of neural networks for signal processing tasks including time series prediction and system identification. Keywords are; networks architectures, preprocessing, network training, validation and generalization.
|
Contents: Vector quantization with application to speech technology: Steffen Duus Hansen.
The purpose is to provide the participants with the necessary knowledge in order to design vector quantizers for different speech technology applications. Keywords are; the fundamentals of vector quantization, low rate speech coding (2-10 kbit/sec.), Hidden Markov Models, and speech recognition.
Adaptive signal processing, filter banks and wavelets: John Aasted Sørensen. The objective is to provide the participants with algorithms which constitute fundamental building blocks in adaptive signal analysis, separation, compression, and expansion. Keywords are; the stable fast transversal FIR filter (SFTF), singular value decomposition (SVD) and multiple signal classification (MUSIC) for signal separation and noise reduction, filter banks and wavelets for analysis and synthesis.
|
|
|
