Skip to Main content Skip to Navigation
Conference papers

Modélisation d'un contexte global d'étiquettes pour l'étiquetage de séquences dans les réseaux neuronaux récurrents

Abstract : During the last few years Recurrent Neural Networks (RNN) have reached state-of-the-art performances on most sequence modeling problems. In particular the sequence to sequence model and the neural CRF have proved very effective on this class of problems. In this paper we propose an alternative RNN for sequence labelling, based on label embeddings and memory networks, which makes possible to take arbitrary long contexts into account. Our results are better than those of state-of-the-art models in most cases, and close to them in all cases. Moreover, our solution is simpler than the best models in the literature. MOTS-CLÉS : Réseaux neuronaux récurrents, contexte global, Étiquetage de séquences.
Complete list of metadata

Cited literature [31 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02002111
Contributor : Loïc Grobol <>
Submitted on : Thursday, January 31, 2019 - 3:34:50 PM
Last modification on : Tuesday, January 5, 2021 - 5:28:07 PM
Long-term archiving on: : Wednesday, May 1, 2019 - 6:58:16 PM

File

paper1.pdf
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution 4.0 International License

Identifiers

  • HAL Id : hal-02002111, version 1

Collections

Citation

Marco Dinarelli, Loïc Grobol. Modélisation d'un contexte global d'étiquettes pour l'étiquetage de séquences dans les réseaux neuronaux récurrents. Journée commune AFIA-ATALA sur le Traitement Automatique des Langues et l’Intelligence Artificielle pendant la onzième édition de la plate-forme Intelligence Artificielle (PFIA 2018), Jul 2018, Nancy, France. ⟨hal-02002111⟩

Share

Metrics

Record views

147

Files downloads

126