EE 225A Spring 2005

Announcements

 

Please read these announcements frequently. Hit the 'refresh' button to make sure you are viewing the latest version.

 

---

Homework mentioned below (and solutions) is available in the following directory [dir]

Other documents (like M-files used in lecture) are available in the following directory [dir]

---

May 15

Grades have been submitted to BearFacts (at midnight Sunday). The class GPA is 3.54 (the Department guideline is 3.4+-0.2). I am not sure of the latency until your grade becomes available to you, but assume it is less than 24 hours. Have a nice summer!

May 13

Second exam grades and course averages have been posted.

May 4

This outline of the topics in the course will be covered in class tomorrow. The derivation of the matrix inversion lemma in class yesterday had a flaw. Here is a different proof.

May 3

After your group has turned in your project report, each student in the group needs to individually send me email with the 100 points allocated among your fellow group members. (This will be in the form of a list of the names of your other group members, with an allocation of points to each member, adding to 100 total.) I will keep your allocation private, informing each student on the total points they received, and will not announce your group project grade until each of you has sent me your allocations. Here is the description from the project page: "An individual grade (20%) determined by peer evaluation, based on each student's contribution to the overall project outcome. Each student is allocated 100 points total, and asked to anonymously divide these points among the other group members based on the quality and quantity of their contribution to the project outcome. For example, if the group has four members, so there are three 'other' members, and a student deems the contributions of the other three members to be equivalent, then s/he would allocate 33.3 points to each of the other students. If s/he deemed one student to have made twice the contribution of the others, then s/he would allocate the points as 50, 25, and 25."

May 2

Reminder:  Project reports due May 3, and second midterm is on May 10.

 

Office hours before the exam:  May 5 (4-5pm), May 9 (3-5pm), May 10 (11-12am).

 

Readings for this week:

×         For May 3, read Hayes Sections 8.6.1 and 9.4.1

×         May 5 we will review the major themes of the course and do the HKN teaching evaluations; there will be limited time for questions.

April 22

To help you study for the second midterm, I posted solutions to some of the most relevant book exercises in Chapters 7, 8, and 9 in the file "Soln10.pdf". I suggest that you go through the solutions, or even better try out the problems first.

April 20

A reminder: second midterm exam is on Tues May 10 in 258 Dwinelle, 2-3:30pm. It will emphasize the material in the second half of the course: adaptive filters, optimum filters, frequency-domain representations, and applications. However, it should also be recognized that the course is cumulative (the second half of the course builds on the first) and one goal of the exam is to see how well you can integrate the concepts across the entire course. The best guideline for what concepts will be emphasized is what has been covered in lecture. The exam will have two parts: Part I: Approximately 40 min of multiple choice questions (closed book and notes) and Part II: 40 min of problem solving (open book and notes). (You get to choose what time you shift from Part I to Part II.) Please bring a blue book (these are stapled pages with a blue cover specifically for exam taking available at the bookstore) for Part II.

April 17

Readings for this week:

§         For April 19, Hayes sections 8.1-8.2 focusing on the filter bank interpretation.

§         For April 21, Hayes section 8.2, focusing on the periodogram interpretation. We won't cover the detailed methods in class (Bartlett, Welch, Blackman-Tukey) but rather be satisfied with conveying the basic ideas.

§         Review Hayes section 8.5 this week, which should already be quite familiar.

 

On April 26, Prof. Edward Lee will give a guest lecture on speech processing. This will be a good illustration of the filter bank and spectrum estimation techniques.

 

A reminder that class is cancelled on April 28.

 

No homework this week: work on your projects! Remember that the project report deadline is May 3—it is coming upon us quickly!

April 15

Midterm #1, Part I, #8: A student pointed out that the fourth item should not be checked (there will always be solutions, since the Grammian matrix must be non-singular). I have corrected the posted solutions, and added additional explanation. If you did not check the fourth item, and had 1/5 points deducted as a result, please inform me by email and I will give you a point back. If you did check the fourth item, no need to do anything.

 

Homework #9 solutions: a student pointed out an algebraic error, which has been corrected.

April 12

Readings for April 14:

§         Review Hayes section 7.2 (we have covered most of this already)

§         Read Hayes section 7.3.

§         Read Hayes section 7.4. Note that we will not cover 7.4 in its entirety, but merely demonstrate a Kalman filter for an easy special case and the general structure of the problem and the solution. You will only be responsible for the low level of detail covered in class.

April 2

Readings before lecture this week:

§         On April 5, we will continue with adaptive filtering applications, and then start looking at other ways to accommodate non-stationary signals. Please review the autocorrelation and covariance methods in Section 4.6 and the recursive least-squares methods in 9.4 (we will not derive the algorithms at this time, but just point out the methodology).

§         On April 7 and 12, we will revisit the lattice filter, and look at ways to adapt it. Please review Sections 7.2.4, 6.1, 6.2, 6.5.4, (we will not cover the other subsections of Section 6.5 in class), 9.2.7-8.

March 31

If you use Hayes’ lms.m to run the adaptive filter convergence example given in lecture today, you have to replace for “k=2:M” with “for k=2:M-nord+1” to make it work.

March 30

Homework #9 is due April 7

 

A reminder that project Milestone 2 is due Thursday at midnight. This milestone is not graded, but the purpose is to elicit useful feedback on the topic and scope of your project, helping to make it a more educational experience.

March 27

Midterm has been completely graded; see grade postings [html]; report any errors to the instructor. Solutions have also been posted in the homework solutions directory.

March 22

Reading assignments in advance of class:

March 14

Homework #8 is due March 31

 

A reminder: midterm exam is on Thur March 17. It will cover through lattice filters (Hayes Chapter 5) and Homework #7, and the best guideline for what concepts will be emphasized is what has been covered in lecture. The exam will have two parts: approximately 40 min of multiple choice questions (closed book and notes) and 40 min of problem solving (open book and notes). (You get to choose what time you shift from multiple choice to problem solving.) Please bring a blue book (these are stapled pages with a blue cover specifically for exam taking available at the bookstore—buy extra for the second midterm).

March 8

A reminder that starting today, Tuesday office hours have been moved from 4-5pm to 11-12am. On Thursdays, office hours will remain 4-5pm.

March 7

Starting on March 10, we will spend three lectures on adaptive filtering (I think we have enough background at this point, it is a practically important topic, plus this background will be very useful for the projects). Advance readings:

 

No new homework assigned this week so you can focus on the midterm.

 

Advance notice: Class and office hours will be canceled on April 28 due to the EECS Faculty Retreat.

March 3

As there seemed to be quite a few questions about the project format, a more detailed project description has been posted [html].

March 2

I added a couple illustrations to the singular matrix writeup posted yesterday [pdf].

 

Homework #7 is due March 10.

March 1

I more carefully worked the solution to the 2x2 singular matrix case that we were confused about in class [pdf]. Let me know if you have any questions about this—I hope this will avoid using any more class time on this example!

Feb 26

The section on lattice filters in "Geometric interpretation of signals: applications" [pdf] (pages 7-on) has been updated and expanded.

 

This week we will be covering the all-zero (FIR) lattice filter. Readings in advance of March 1:

 

Readings in advance of March 3:

Feb 24

The Matlab program demoed in class had an error that was causing non-conjugate poles-zero pairs to be plotted. The file has been corrected so the model pole-zero plot should now be correct.

Feb 19

In homework #5, Hayes problem 3.4 should not have been assigned since it is a duplicate of homework #4. I hope you didn’t do it twice J.

 

Starting this week we are going to be actively using linear algebra, so you should be more seriously reviewing section 2.3. In particular, we will review section 2.3.6 in some depth, although we will use the projection theorem to derive and interpret the results.

Feb 16

Repaired a few typos in Homework #5 and posted the new version at 2:30pm today.

 

It is time to ask for your feedback on how the course is going. If you would like to provide suggestions for improvement, please print out this form, fill out, and return to the instructor in class.

Feb 15

After class a couple people seemed confused about how to relate a power spectrum at the input and output of a filter. See Eq. 3.92 on page 101.

 

Readings in advance of Feb 17:

 

Readings in advance of Feb. 22:

 

Readings in advance of Feb 24:

 

Recall that project milestone 1 (form a group of three students and at least two applications to run by the instructor for comment) is due March 8. It is a good time to get started on this.

Feb 10

We stated in class that the MMSE conditional-mean estimator for jointly Gaussian RV’s is linear. A small Matlab program in [dir] calculates the exponent of the conditional density -- it is obviously Gaussian with a mean that is linear in y. You can even verify the linear coefficient.

Feb 9

The first example on p.9 in "Geometric interpretation of signals: background" [pdf] was updated. A new Version 1.3 was posted today with this change.

 

In preparation for Feb 10:

 

Readings in advance of Feb 15:

Feb 3

Readings in advance of Feb 8:

Readings in advance of Feb 10:

Jan 27

Readings in advance of Feb 1:

Readings in advance of Feb 3:

Jan 24

Readings in advance of Jan 25:

Readings in advance of Jan 27:

Jan 20

Version 1.1 of “Real functions of a complex variable” has been posted, adding an example and correcting a few typos [pdf].

Jan 18

In this class you will be expected to do some Matlab programming in this class. You should obtain Matlab (a student version is available at nominal cost) or if you prefer you can obtain an account on the EECS instructional computer cluster. Please begin honing up on Matlab programming if you are not already familiar with it. Hayes has a number of Matlab programming examples in the book, and these are described systematically in the Appendix.

 

For each lecture, read the background material from the book before class. The class discussion will assume you have read that material. These before-class readings will be posted here.

 

Readings and homework for Jan 18:

 

Readings and homework for Jan 20: