EE 381K Information Theory - Fall 2025

• Meets TuTh 9:30-11

Instructor

• Gustavo de Veciana
• Office: EER 6.874
• Office Hours: Please see my home page
• Email: gustavo@ece.utexas.edu
• WWW: http://ece.utexas.edu/~gustavo

Description

This course is intended for graduate students with a background in probability and ideally some background/interest in communications, systems, signal processing and/or data sciences. For details see Canvas web page.

Course Topics (Tentative list)

• Entropy, Relative Entropy and Mutual Information: Definitions: chain rules for entropy, Jensen's inequality and its implications, Data Processing Inequality, Thermodynamics, Fano's inequality.
• The Asymptotic Equipartition Property: AEP, Consequences of AEP and data compression, high probablity sets and typical set
• Entropy Rates of stochastic processes Markov chains, entropy rate, Hidden Markov Models.
• Data Compression : Codes, Kraft inequality, Huffman codes, Arithmetic codes,
• Gambling and data compression Gambling with side information, dependent horse races , data compression and gambling, entropy of english
• Kolmogorov Complexity: Models of computation, Kolmogorov complexity, The halting problem and non-computability of Kolmogorov complexity Occams Razor..
• Channel Capacity: Symmetric channel, Jointly typical sequences, Channel Coding Theorem, Joint source channel coding theorem
• Differential Entropy: AEP for continuous RVs, differential entropy vs discrte entropy, Joint and conditional differential etnropy, relative entropy and mutual information
• The Gaussian Channel: The Gaussian channel, parallel Gaussian Channels, Channels with colored Gaussian Noise, Gaussian Channels with feedback
• Maximum Entropy and Spectral Estimation: Maximum entropy distribution, spectrum estimation, Burg's maxmum entropy Theorem
• Information theory and statistics : Method of types, LLN, Large deviation theory, Sanov's theorem, Hypothesis testing, Lempel-Ziv Coding
• Rate distortion theory: Quantization, Rate distortion function, Rate Distortion Theorem,
• Network Information theory: Medium Access channel, Broadcast channel, source coding with side information, rate distortion with side information
• Assorted topics we would like to touch on (perhaps through course projects)
  • Machine Learning and Information Theory
  • Quantum Information Theory
  • Image and video compression and quality assessment

Prerequisites

A prerequisite for this course is a graduate course in Probability and Stochastic Processes , such as EE381J.

Required text

Elements of Information Theory , by Cover and Thomas. Wiley, 2006. Second Edition

Format/Evaluation

Homework will be assigned weekly. There will be two in-class midterms. No final exam, but there will be a project presentation on a topic of your choice (subject to my approval) - attendance to peer project presentations is mandatory.

Final Exam:

No final, however you will be required to select a topic (possibly a research topic) and make a presentation to the class. Your time may be prior to or during the scheduled final exams time - participation during your colleagues presentations mandatory.

Where does course this fit in?

Prior to taking this course, you might consider taking: Probability and Stochastic Processes I this is a foundations graduate course and perhaps Digital Communications.. Some related courses are, Wireless Communications; Digital Signal Processing. Advanced Signal Processing.