1st Semester 2013/14: Information Theory: Codebreaking, Gambling, Inference
- Statistical inference from observed data to hidden parameters always involve a degree of uncertainty. Information theory is a mathematical tool that allows to quantify this uncertainty in a meaningful way, and thus to express the limits of what we can learn from incomplete or noisy data. This has implications for how we understand human learning and communication, and how we should approach engineering problems like data compression, design of digital communication protocols, etc.
This course will introduce the central concepts of information theory and discuss their interpretation in contexts like forecasting, gambling, and modeling. The course can be taken in tandem with Christian Schaffner's information theory course, which starts immediately after.
- The course will run from January 13 to the last week of January (the exact date depending on the number of participants). The first five days will consist of a crash course in information theory. The second week is reserved for preparation of the student presentations, and the third week for giving those presentations.
- The course is intended to be accessible for everyone, but a prior knowledge of probability theory will help.
- Students will be required to give presentations in the third week of the course. A list of suggested topics with relevant literature will be provided.
- Website: Slides and other materials will be uploaded to http://informationtheory.weebly.com as they become available.