Scope and Objectives
The objective of this workshop is to advance the current knowledge in nonlinear system identification by encouraging the exchange of ideas and the establishment of formal collaborations between the systems and control, mechanical and machine learning communities.
These three communities have developed over the years various and numerous nonlinear modeling approaches driven by the different backgrounds, constraints and end-uses. Moreover, they generally focus on different aspects of the modeling problem as they face different limiting factors in terms of model quality and identification cost. This is why we believe that, by promoting interaction, significant benefit can be mutually gained.
This workshop will be structured around the benchmark systems available through this website. This website hosts both established benchmarks and recently introduced benchmarks featuring state-of-the-art challenges in nonlinear system identification, such as dynamic nonlinearity, process noise, and high order dynamics in combination with a hard nonlinearity.
Solicited contributions should describe solutions or analyses to one or several of the benchmark problems. In particular, comparative overviews of methods would be particularly appreciated.
A hands-on mini-course in nonlinear system identification will take place from Monday 09/04/2018 to Tuesday Noon 10/04/2018, right before the 2018 Benchmark Workshop. This mini-course will cover both parametric and nonparametric identification, linear, block-oriented and nonlinear state-space model structures. A hands-on approach is maintained throughout the course, participants will be able to apply the covered topics to one of the benchmark examples. A preliminary overview of the mini-course can be found below:
(1) An Introduction to Nonlinear System Identification by Johan Schoukens (Vrije Universiteit Brussel, Belgium)
(2) An Introduction to Bayesian versus Maximum Likelihood Estimation by Koen Tiels (Uppsala University, Sweden)
(3) Identification of Block-Oriented Nonlinear Systems by Maarten Schoukens (TU Eindhoven, The Netherlands)
(4) Identification of Nonlinear State-Space Systems by Jean-Philippe Noël (University of Liège, Belgium)
(5) Hands-On sessions on the Best Linear Approximation, Multisine Signals, Block-Oriented and Nonlinear State Space Identification by Jean-Philippe Noël, Maarten Schoukens and Koen Tiels
The mini-course is open to both workshop participants, and researchers who did not attend the 2018 Benchmark Workshop. Registration for the mini-course is to be completed by January 31, 2018 through this link.
The workshop takes place from Wednesday morning 11/04/2018 to Friday noon 13/04/2018. An overview of the workshop program will be announced later.
We are proud to announce the following keynote speakers:
(1) Peter Young, University of Lancaster, UK
(2) Johan Suykens, KU Leuven, Belgium
(3) Alfred Schouten TU Delft, The Netherlands
(4) Eleni Chatzi, ETH Zürich, Switzerland
(5) Jorge Goncalves, University of Luxembourg, Luxembourg
Participant Registration and Deadlines
Registration for the workshop is to be completed by February 14, 2018 through this link. Late registration is still possible by contacting the organizers. Presenting participants are also requested to submit a 1-page abstract summarizing the main aspects of their contribution by March 18, 2018. A registration fee of 100 euros is requested to participate in the workshop.
Please email Maarten Schoukens (email@example.com) or Jean-Philippe Noël (firstname.lastname@example.org) to submit your abstract, or ask any other questions.
(1) Registration deadline: Wednesday February 14, 2018, register here
(2) Abstract submission deadline: Sunday March 18, 2018 - 1-page abstract template available here
(3) Notification of acceptance: Sunday March 25, 2018
When and Where
Mini-Course Dates: Monday 09/04/2018 - Tuesday Noon 10/04/2018
Workshop Dates: Wednesday 11/04/2018 - Friday 13/04/2018
Venue: ULg, Campus Sart-Tilman, Liege, Belgium