Näitä sivuja ei päivitetä enää. Ole hyvä ja katso tietojenkäsittelytieteen laitoksen WWW-sivuja: http://ics.tkk.fi/fi/studies/.
These pages are not any more updated. Please, see web pages of Department of Information and Computer Science (ICS): http://ics.tkk.fi/en/studies/.
|Notices (updated 30.11.2006)
o 30.11. Added the exercise of last meeting on 28.11.
o Return the final exercise to the course assistant by 12.12.
o Thanks to all those who attended the course!
|Lecturer||Professor Samuel Kaski, Laboratory of Computer and Information Science|
|Assistant||M.Sc.(Tech.) Jarkko Salojärvi, Laboratory of Computer and Information Science|
|Semester||Autumn 2006, periods I and II|
|Seminar sessions|| On Tuesdays at 12-14 in lecture room A328 in computer science
Konemiehentie 2, Otaniemi, Espoo. The first session on Thursday 14.9.2005.
|Registration||Send mail to firstname.lastname@example.org|
The goal is to read and understand the book Bayesian Theory (Bernardo & Smith, Wiley 2000). The book starts by viewing Bayesian inference as a special case of decision theory, and then proceeds by giving a comprehensive discussion on modelling and inference. The focus will be more on the latter parts, but we will go through the beginning too, most likely skipping the proofs there.
We will all read a chapter at a time, and discuss it during the next session. One of the participants, each in turn, introduces the chapter, leads the discussion, and coordinates exercises with the help of the course staff.
Basic math courses of TKK or equivalent, and reasonable knowledge of a subset of statistics, probabilistic modeling, machine learning, pattern recognition or information theory. The more theoretical T-61 courses give a very suitable background, for instance.
Reading cycle (introduction of a chapter and leading the discussion + active participation + summaries + exercises) (7 cr).
A combination of exercises, summaries and active participation produces 7cu in the form of T-61.6900 Yksilölliset opinnot. The course is graded as fail/pass/pass with distinction. The grade depends on the number of exercises solved, and active participation in the discussion.
Send a short description by email to the course assistant (email@example.com), describing your reason for joining the reading circle, plus your potential contribution to the circle. Because of the format, we cannot take more than 15 students, but otherwise there is no need to be formal. More advanced researchers are welcome as well. Deadline: 12.9.
We will first go through the exercises given in the previous meeting. Those who want credits need to return their solutions to the assistant before the meeting. In principle everyone can present the solutions, but at least the one that gave the exercises should have the model solution.
We will then have a presentation, interleaved with discussion. In the end the presenter will give 1-3 exercises (for those wishing to get credits) on the subject. This will take 2 hours in total.
Note: The course is a very informal reading circle. The actual format is much up to the participants, except if you want to get credits. Each presenter can choose their own presentation style, most important is to understand and discuss. We expect everybody to read the book, there is no use just to listen. The focus is not on the proofs (but you can do that too), but instead on understanding why, and what are the implications.
Discussion related to practical applications and implications to our own research is favoured, and even philosophical discussion is quite ok (let's restrict the time spent on that, though). Naive questions are strongly solicited, especially if you did try to understand the issue!
Our plan is to read chapters 2-3 relatively quickly, and then read the rest. Finally, we will get back to the beginning.
|Time||Lecturer||Subject and material|
For more information, please send email to firstname.lastname@example.org.
Samuel Kaski and Jarkko Salojärvi
You are at: CIS → T-61.6900 Individual studies: Reading circle on Bayesian theory
Page maintained by email@example.com, last updated Thu Nov 30 12:19:52 EET 2006