Hybrid Language Processing in the Deep Learning Era
Résumé
In this talk we examine critically the current wave of interest in pure deep learning for natural language processing. What can symbolic resources do for natural language processing? Among other examples, we take into account the languages with more restricted graphical delimitation than English. Then we discuss the foreseeable future of the synergy between machine learning and symbolic resources: are the goals of formalisation, precision, reliability, adaptability within reach for linguistic data?
Domaines
Informatique et langage [cs.CL]
Origine : Fichiers produits par l'(les) auteur(s)