Nonlinear Black Box Models in System Identification
1997 (English)In: Proceedings of the 1997 IFAC International Symposum on Advanced Control of Chemical Processes, 1997, 1-13 p.Conference paper (Refereed)
The basic idea behind the large family of non-linear black box models is described. It is shown how neural networks, wavelet networks, hinging hyperplanes, kernel methods, fuzzy models, etc all fit to a common framework of basis expansion, using a single 'motherfunction'. Most differences relate to how this single function is expanded to higher regressor dimensions. The estimation theory, algorithmic aspects and applications to dynamical systems are also covered.
Place, publisher, year, edition, pages
1997. 1-13 p.
Nonlinear black box, Neural networks, Kernel methods, Dynamical systems
IdentifiersURN: urn:nbn:se:liu:diva-91579OAI: oai:DiVA.org:liu-91579DiVA: diva2:626299
1997 IFAC International Symposum on Advanced Control of Chemical Processes, Banff, Alberta, Canada, June, 1997