PAC-Learning and Asymptotic System Identification Theory
1996 (English)Report (Other academic)
In this paper we discuss PAC-learning of functions from a traditional system identification perspective. The well established asymptotic theory for the identified models' properties is reviewed from the PAC-learning perspective. The role of finite-dimensional, smooth parametrizations over compact parameter sets is spelled out. This also sets some limits for the interest of identification-theory type results in a learning-theory context.
Place, publisher, year, edition, pages
Linköping: Linköping University Electronic Press, 1996. , 5 p.
LiTH-ISY-R, ISSN 1400-3902 ; 1884
Learning theory, PAC learning
Cybernetik Informationsteori, Maskinelement Servomekanismer Automation
IdentifiersURN: urn:nbn:se:liu:diva-55316ISRN: LiTH-ISY-R-1884OAI: oai:DiVA.org:liu-55316DiVA: diva2:316068