My licentiate thesis Hidden Markov models: Identification, control and inverse filtering is now written and has been printed. If you want a physical copy, pass by my office and I will be glad to give you one. It is also available online here.
The opponent for my licentiate seminar will be Prof. Eric Moulines of Ecole Polytechnique in Paris, France.
If you have any comments and/or suggestions, please feel free to contact me.
I will be teaching the course EL2800 Stochastic Control and Optimization together with Alexandre Proutiere and Vahan Petrosyan this period.
The notes from my exercise-sessions can be found on the Teaching page.
I attended this year’s ERNSI workshop and presented a poster related to our NIPS paper on Inverse filtering for hidden Markov models. The European Research Network for System Identification (ERNSI) workshop is a yearly workshop that was launched in 1992. The workshop was this year held in Lyon, France.
I am very glad to be able to announce that out of the 3240 submissions that were made, our paper Inverse Filtering for Hidden Markov Models is one of the 678 that were accepted for presentation at the 31st Annual Conference on Neural Information Processing Systems (NIPS). This year, the conference is held December 4-9 in the Long Beach Convention Center, Long Beach, USA.
The paper is joint work between me, Cristian Rojas, Vikram Krishnamurthy and Bo Wahlberg. The paper introduces and provides initial solutions to a new class of inference problems. Potential areas of applications include, e.g., machine learning, signal processing, fault detection, microeconomics and finance.
I attended the PhD course on Sequential Monte Carlo methods, given by Thomas Schön and Fredrik Lindsten of the Department of Information Technology at Uppsala University.
I visited Vikram Krishnamurthy at the newly built campus of Cornell Tech on Roosevelt Island, Manhattan, together with Bo Wahlberg. I gave a seminar on Inverse filtering in hidden Markov models – a topic we have been working on lately.
The paper Identification of Hidden Markov Models Using Spectral Learning with Likelihood Maximization has been accepted for presentation at, and publication in the proceedings of, the 56th IEEE Conference on Decision and Control, to be held in Melbourne, Australia on December 12-15, 2017.
This summer, I attended the VUB-ELEC’s 2017 Workshop on System Identification.
The following courses were lectured: Frequency Response Function Measurements (Non-Parametric Tools) by John Lataire, Yves Rolain and Rik Pintelon; Dynamic System Identification (Parametric Tools) by Rik Pintelon, Ivan Markovsky, Philippe Dreesen, John Lataire, Maarten Schoukens and Koen Tiels; and Control-Oriented System Identification by Simone Formentin.
Our paper Computing monotone policies for Markov decision processes: a nearly-isotonic penalty approach is now available on arXiv. It will be presented at the 20th World Congress of the International Federation of Automatic Control (IFAC WC’17) that will be held July 9-14 2017 in Toulouse, France.
Our new work An Optimal Gender-Specific Treatment Policy for Abdominal Aortic Aneurysms Constructed Using a Markov Decision Process Model has been accepted for presentation at the Society for Vascular Surgery’s 2017 Vascular Annual Meeting, to be held in San Diego, California from May 31 to June 3, 2017.
The work has been developed jointly with Antti Siika and Joy Roy at Karolinska Institute. It is a continuation of our previous article A Markov Decision Process Model to Guide Treatment of Abdominal Aortic Aneurysms.
This is a personal web page. Opinions expressed are solely my own and do not express the views or opinions of my employer.
