ELG5377 Adaptive Signal Processing, Fall 2014
Contents
- To promote the use of the discussion forum "Exercise solutions" on the
Blackboard Learn site, 5 of the 20 marks for 'Assignments' will be awarded
for participation. At the end of the course, I will evaluate each student's
level of participation in this activity to assign a mark out of 5.
- Midterm date changed to Oct. 22, 8:30 am in the usual classroom.
- Lecture 1: Introduction (PDF)
- Lectures 2,3: Vector spaces (PDF) updated
2014-09-10
- Lectures 4, 5, 6, 7, 8: Stochastic processes (PDF); random
variable example overhead (PDF)
- Lecture 5, 9: Introduction to Wiener filtering (PDF); updated 2014-10-01 (PDF)
- Lecture 7, 8, 9: Wiener filter example (PDF); updated 2014-09-29 (PDF); further updated
2014-10-01 (PDF)
- Lecture 10, 11, 12: Linear prediction (PDF); updated 2014-10-27 (PDF)
- Remaining lecture slides will only be posted on Blackboard Learn.
- A. Sayed, Adaptive
Filters, Wiley-IEEE Press, 2008. Available as ebook
through University of Ottawa library.
- Sept. 4: Introduction to the course. Slides are posted.
- Sept. 8: Vectors spaces, sections 1-10 of the course notes. You are in a
position to work on exercises 1-12 in the notes on vector spaces.
- Sept. 10: Vector spaces, section 11 of the course notes, as well as
numerical example in the slides. Introduction to notions of probability.
Exercises 13-15 can now be attempted.
- Sept. 15: Review of notion of random variable, stochastic process,
moments, ergodicity, LSI filtering of WSS processes, power density
spectrum. Slides 1-16. Related material in Haykin 5e, sections 1.1, 1.2,
1.12, 1.13, 1.14.
- Sept. 17: Stochastic processes: Cross-correlation, filtering of WSS
processes (slides 17-24). Presentation of Term Project. Introduction to
Wiener filtering (slides 1-7). Related material in Haykin 5e, sections 2.1
to 2.5.
- Sept. 22: Finished the introduction to Wiener filtering. Covered
bandlimited processes, harmonic processes, the general linear process, and
autoregressive processes (slides 25-43). Related material in Haykin 5e can
be found in section 1.5 on stochastic models and sections 1.7 to 1.9 on
autoregressive models.
- Sept. 24: Finished AR process, stochastic processes slides 44-50. Went
through detailed Wiener Filter example. Slides posted. Essentially all the
material for the exercises on stochastic processes has been covered (except
Haykin #10).
- Sept. 29: Finished Stochastic Processes: MA and ARMA processes.
Simulation of AR processes and Wiener filtering. Comparison of time
averages and ensemble averages. Slides updated and MATLAB files posted on
Blackboard.
- Oct. 1: IIR Wiener filter for general linear process. Wiener filter
slides updated. Continuation
of Wiener filter example. Slides updated. Introduction to
eignenvalue and eigenvector analysis of correlation matrix. You are now in
a position to complete all the exercises on Wiener filters except part (c)
of the first two questions (from Haykin). This will be completed in the
next lecture.
- Oct. 6: Completed Wiener filter using eigenvector basis, finished Wiener
filter slides. Started Linear Prediction: FIR and IIR MMSE linear
prediction. Introduction to "Backward Linear Prediction". Slides 1-18.
- Oct. 8: Backward linear prediction, Gram Schmidt orthogonalization,
Levinson algorithm. Slides 17-40.
- Oct. 13, 15: Study break.
- Oct. 20: Prediction error filters, the lattice structure, joint-process
estimation. Slides 41-56.
- Oct. 22: Midterm exam.
- Remaining summary of lecture contents will be posted on Blackboard Learn,
in the section on Lecture Slides.
- Assignment 1 description and solution on Blackboard Learn.