.

University of Washington, Department of Electrical Engineering

 EE 595A: Information Theory, Winter 2004 

Instructor

Information

Announcements

Homework

Links


Instructor


Prof.
Jeff A. Bilmes (bilmes@ee.washington.edu)
Office: 418 EE/CS Bldg., 221-5236
Office hours: Fridays, 11:00am-12:30pm, 418 EE/CS Bldg.


Information

This course will be an thorough introduction to information theory.

Topics will include entropy, mutual information, asymptotic equipartition properties, data compression to the entropy limit (source coding), communication at the channel capacity limit (channel coding theorem), coding theory (including ECC and other modern codes), method of types, differential entropy, maximum entropy, rate-distortion theory, alternating minimization, and information geometry in general. Additional topics will include information theory as it is applicable to pattern recognition, natural language processing, computer science and complexity, biological science, and communications.


Prerequisites: basic probability, statistics, and random processes (e.g., EE505 or a Stat 5xx class or consent of the instructor). Knowledge of matlab. The course is open to students in all UW departments.

Course Format: Four hours of lecture (MW 3:30-5:20 EE1- 042) per week. Final exam 2:30-4:20 p.m. Thursday, Mar. 21, 2002

Texts: We will use two texts including:

  1. A classic by Thomas Cover and Joy Thomas entitled Elements of Information Theory (1991), Wiley.
  2. The new text by David Mackay from Cambridge University, entitled Information Theory, Inference, and Learning Algorithms, 2003, Cambridge.

Course overview in PS and PDF formats, which gives more information (such as grading policy, other interesting texts, etc.).


Announcements


(Feb 10th) A
practice midterm is now available from which to study.

(Feb 9th) The figure from today's lecture is online

(Jan 5th) Please fill out the following survey (PS or PDF) if you haven't already. .


Homework

Homework 1, due Wednesday, Jan 21st.
PS or PDF (solutions in PS and PDF)

Homework 2, due Wednesday, Feb 4th, in class. PS or PDF (solutions in PS and PDF)

Homework 3, due Friday, Feb 27th, at 4:30pm. PS or PDF (solutions in PS and PDF)

Homework 4, due Wednesday, March 10th, in class. PS or PDF (solutions in PS and PDF)


Links

Claude Shannon's obituary, Feb 24th, 2001

Course news group (it is listed as ee595, but this is our course news group, the last time the course was offered, it was ee595).

Shannon's original paper

IEEE Information Society Home Page

IEEE Electronic Library and access to online version IEEE Transactions on Information Theory for UW locations.

Entropy on the WWW

Arithmetic coding for data compression, an article written in 1987 that has become the standard introduction on this topic.

Arithmetic coding revisited Written by the same authors as the previous article, this one written in 1998 provides an update to issues surrounding arithmetic coding.

Data compression, a survey of the field to 1987.

Bayesian networks for lossless dataset compression. Can Bayesian networks be used for compression? Read here and find out.

Modeling for text compression, a good survey on text compression. The same authors later wrote a book on the subject with the same title.

High Quality Document Image Compression with DjVu. by Léon Bottou, Patrick Haffner, Paul G. Howard, Patrice Simard, Yoshua Bengio, and Yann Le Cun, describes the DjVu method in detail.

Is Huffman coding dead?, with a title like this, you have to take a look.

The Miraculous Universal Distribution, Ming Li's introductory essay about Kolmogorov complexity and randomness. Also see Paul Vitanyi's extended randomness paper.

Quotations by Laplace about Kolmogorov complexity, but 100 years earlier.

David MacKay's online IT course



Maintained by Jeff Bilmes Last updated: $Date: 2004/03/21 07:07:21 $