Learning typed automata from automatically labeled data

Kermorvant, Christian;de la Higuera, Colin;Dupont, Pierre
(2004) Journal electonique d’intelligence artificielle — Vol. 6, n° 45 (2004)

Files

LEARNINGAUTOMATA.pdf
  • Restricted Access
  • Adobe PDF
  • 273.38 KB
Request a copy

Details

Authors
  • Kermorvant, ChristianUniversité de Montréal, Canada
    Author
  • de la Higuera, ColinUniversité Jean Monnet, Saint-Etienne, France
    Author
  • Dupont, PierreUCLouvain
    Author
Abstract
In this paper, we propose a way of incorporating additional knowledge in probabilistic automata inference, by using typed automata. We compare two kinds of knowledge that are introduced into the learning algorithms. A statistical clustering algorithm and a part-of-speech tagger are used to label the data according to statistical or syntactic information automatically obtained from the data. The labeled data is then used to infer correctly typed automata. Compared to a previously proposed method which improved grammatical inference, our approach yields better automata. The inference of typed automata with statistically labeled data provides language models competitive with state-of-the art n-grams on the Air Travel Information System (ATIS) task.
Affiliations

Citations