News

December 22, 2015; We have got a paper accepted for publication at the 19th International Conference on Artificial Intelligence and Statistics (AISTATS) in Cadiz, Spain in May. The paper is titled Computationally efficient learning of Gaussian process state space models together with Arno Solin, Simo Särkkä and Thomas Schön (see more details, including preprint, on this page).

In short, the problem addressed is learning/identification of a state space model where a Gaussian process prior is assumed for the unknown transition function for the (unobserved) states from time t to time t+1. Particle filters play an important role in the proposed algorithm.

September 16, 2015: We got two papers accepted for publication at the 6th IEEE international workshop on computational advances in multi-sensor adaptive processing (CAMSAP) in Cancun, Mexico in December.

The first paper, Nonlinear State Space Model Identification Using a Regularized Basis Function Expansion (with Thomas Schön, Arno Solin and Simo Särkkä), deals with black box identification of nonlinear state space models, using particle filter-based methods for learning the weights in a basis function expansion inside a state space model. A related work is also covered by Computationally Efficient Bayesian Learning of Gaussian Process State Space Models (currently at arXiv), which is focusing on Bayesian learning of the same type of model. I also have a dedicated page for this work under construction.

The second paper, Marginalizing Gaussian Process Hyperparameters using Sequential Monte Carlo (with Johan Dahlin and Thomas Schön), presents how to apply a SMC sampler to marginalize hyperparameters in Gaussian process regression.

May 31, 2015: I got the paper Nonlinear state space smoothing using the conditional particle (together with Thomas Schön and Manon Kok) accepted for publication at 17th IFAC Symposium on System Identification (SYSID 2015), Beijing, China, October 19-21. A pre-print of the submitted version is available at arXiv, and Matlab code here.

The tutorial paper Sequential Monte Carlo Methods for System Identification written by our team was also accepted for the same conference, also with a pre-print availabe on arXiv.

January 16, 2015: The site for our racetrack, CARS, is finally online! Please have a look at cars.it.uu.se. All students involved in the projects have done a fantastic work!

June 21, 2014: My paper Identification of jump Markov linear models using particle filters (together with Thomas Schön and Fredrik Lindsten) has been accepted for presentation at the 53rd IEEE Conference on Decision and Control, Los Angeles, California, USA, December 15-17, 2014. The paper can be found here.

May 22, 2014: A popular science description of my research is now available (in Swedish)!

March 27, 2014: The Matlab code for my recently submitted publication is available on this page.

October 08, 2013: Click on the picture below to see my animation "Particle Filter Explained without Equations"

The Matlab code can be found on this page. Any comments are, of course, very welcome!