MAT-55216 Topics in Applied Mathematics, 5 credits

Spring 2011: Bayesian estimation of time-varying processes: discrete-time systems

Teacher: Dr. Tech. Simo Särkkä, Aalto University, Finland

Coordinator: Prof. Robert Piché, Mathematics Dept, TUT

Schedule: Lectures & Exercises Thursdays 1-4pm (20.1, 3.2, 10.2, 17.2, 24.2, 10.3, 17.3). The exercise hour is 1-2pm and the lecture is 2-4pm. Location: Tampere University of Technology room Td308 (bring a laptop with Matlab/Octave)

Topics: Modeling of time-varying systems with uncertainty, optimal filtering and optimal smoothing, linear and nonlinear Kalman filters (EKF/UKF/SLF/GHKF/CKF), linear and non-linear Rauch-Tung-Striebel smoothers, sequential Monte Carlo methods, particle filters and smoothers. Example applications from navigation, remote surveillance and time series analysis.

Prerequisites: Multivariate calculus, matrix algebra, basic probability and estimation (e.g. one of the TUT courses Bayesian Methods, Stochastic Processes, or Mathematics for Positioning); Matlab/Octave

Course Pass Requirements:

The grade will be based on the project work. See Project work instructions and topics.
Total 2/3 of the home work problems must be completed (see instructions below).

Exercise hour:

Before each lecture there is an exercise hour, which works as follows: In the beginning, you should flag to a list which of the exercises you have completed. One person will be randomly selected to present his/her answer on the drawing board or computer. Note that in the end of the course you must have done and flagged total of 2/3 of all the exercises during the course (14 out of 21 exercises).

The required Matlab/Octave files for the exercises can be found here.

Course Material:

Course booklet: course_booklet_2011.pdf

The slides are copied here after each lecture:

The old slides, booklet and exercises can be found here.

Simo Särkkä

Last modified: Tue Mar 22 10:46:11 EET 2011