Computational Inference in Dynamical Systems

A PhD course at ACFR, Sydney, Australia.

---

General Information

The aim of this course is to provide an introduction to the theory and application of computational methods for inference in dynamical systems. More specifically, the computational methods we are referring to are sequential Monte Carlo (SMC) methods (particle filters and particle smoothers) for nonlinear state inference problems and expectation maximisation (EM) and Markov chain Monte Carlo (MCMC) methods for nonlinear system identification.

A pdf document with a more detailed course description is available here.

Lecture slides are available here.

Contents

• Probabilistic modeling of dynamical systems
• Expectation maximisation (EM) for nonlinear system identification
• Markov chain Monte Carlo methods (MCMC) for nonlinear system identification
• Gibbs sampler
• Metropolis hastings sampler
• Rejection sampling and Importance sampling
• Particle filtering
• Particle smoothing
• Particle MCMC

Organization

This is an intensive course (8 hours of lecturing during 2 days)

• October 17, 2012 (4h), 09.00 - 13.00 in the Mechanical Engineering Tutorial Room 3,
• October 18, 2012 (4h), 09.00 - 11.00 in the AMME Conference Room and 14.00 - 16.00 in the Civil Engineering Lecture Room 2.

Course Literature

Lecture notes will be handed out to the course participants,
[LN] Thomas B. Schön. Computational inference and learning in dynamical systems, 2012.

Prerequisites

Basic undergraduate courses in linear algebra, statistics, signal and systems.

Related Courses

Machine learning, System identification.

Contact Persons

Dr Thomas Schön, tel +46 (0) 13 281373, email: schon_at_isy.liu.se.
Dr Ian Manchester, tel +61 2 9351 2186, email: i.manchester_at_acfr.usyd.edu.au.