EECS 559: Advanced Signal Processing
University of Michigan, Fall
2006
Instructor: Clayton Scott
Classroom: 3433 EECS
Time: MW 10:30-12
Office: 4229 EECS
Email:
Office hours: Friday 10:30-12
Announcements:
- Project description
- Final project poster presentations will take place in EECS 3433 on
Dec. 18, 4-6 PM.
Textbook: Steven Kay, Modern Spectral Estimation, Prentice
Hall
Topics: Spectral estimation (most of course) and
pattern recognition (time permitting).
Prerequisites:
Digital signal processing: LTI systems, filtering, Fourier and
z-transforms
Linear algebra: eigenvalues/vectors, solving linear systems, least
squares problems, projections
Probability: Random variables, expectation, correlation. Familiarity
with random processes and statistical estimation a plus.
Relevant papers, available on IEEE Xplore. To
access
Proc. IEEE, follow Journals and Magazines -> P -> Proceedings of the IEEE.
- Kay and Marple, Spectrum analysis--A modern perspective, Proc.
IEEE, vol. 69, no. 11, 1380-1419, 1981.
- Robinson, A historical perspective of spectrum estimation, Proc.
IEEE, vol. 70, no. 9, 885-907, 1982.
- Johnson, The application of spectral estimation methods to bearing
estimation problems, Proc.
IEEE, vol. 70, no. 9, 1018-1028, 1982.
Lecture notes:
- Overview
- Random processes
- The Power Spectral Density
- The Periodogram
- Classical (Nonparametric) Spectrum
Estimation
- Parametric Modeling: ARMA
Models
- AR Spectral Estimation in Theory
(connections to linear prediction)
- AR Spectral Estimation in Practice
- MA and ARMA Spectral Estimation
- Capon's Method
- Sinusoidal Parametric Modeling
- Bearing Estimation
- Classification
- Plug-In Rules; LDA and QDA
- EM Algorithm for Gaussian Mixture Models
- Kernel classifiers
- Linear classifiers
- Constrained Optimization
- Support vector machines
- Boosting and other voting methods
Homeworks:
- Homework 1, Due 9/13.
- Homework 2, Due 9/20.
Solutions
- Homework 3, Due 9/27.
- Homework 4, Due 10/4.
Solutions
- Homework 5, Due 10/11.
Solutions
- Homework 6, Due 10/25.
Solutions
- Homework 7, Due 11/1.
- Homework 8, Due 11/20.
- Homework 9, Due 11/29.
- Homework 10, Due 12/13.
***Data***
Grading:
- 40% Homework
- 30% Midterm exam
- 30% Final project