On learning equal matrix languages:

Abstract: "We consider the learning problem for languages, called strongly bounded equal matrix languages, consisting of strings of the form [formula] where each a[supscript i] is a symbol and n[subscript i] is a nonnegative integer. The languages are defined in terms of certain parallel rewrit...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: Takada, Yuji (VerfasserIn)
Format: Buch
Sprache:English
Veröffentlicht: Tokyo, Japan 1989
Schriftenreihe:Shin-Sedai-Konpyūta-Gijutsu-Kaihatsu-Kikō <Tōkyō>: ICOT technical report 452
Schlagworte:
Zusammenfassung:Abstract: "We consider the learning problem for languages, called strongly bounded equal matrix languages, consisting of strings of the form [formula] where each a[supscript i] is a symbol and n[subscript i] is a nonnegative integer. The languages are defined in terms of certain parallel rewriting grammars called equal matrix grammars. Also, the languages closely related to semilinear subsets of the Cartesian product of nonnegative integers
We show that (1) the family of strongly bounded equal matrix languages is not learnable from positive examples, while there exists a meaningful subfamily which is learnable from positive examples, (2) given any teacher, called an ideal teacher, the subfamily is learnable in polynomial time of the size of inputs.
Beschreibung:10 S.