Generative Cooperative Networks for Natural Language Generation - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Generative Cooperative Networks for Natural Language Generation

Résumé

Generative Adversarial Networks (GANs) have known a tremendous success for many continuous generation tasks, especially in the field of image generation. However, for discrete outputs such as language, optimizing GANs remains an open problem with many instabilities, as no gradient can be properly back-propagated from the discriminator output to the generator parameters. An alternative is to learn the generator network via reinforcement learning, using the discriminator signal as a reward, but such a technique suffers from moving rewards and vanishing gradient problems. Finally, it often falls short compared to direct maximum-likelihood approaches. In this paper, we introduce Generative Cooperative Networks, in which the discriminator architecture is cooperatively used along with the generation policy to output samples of realistic texts for the task at hand. We give theoretical guarantees of convergence for our approach, and study various efficient decoding schemes to empirically achieve state-of-the-art results in two main NLG tasks.
Fichier principal
Vignette du fichier
lamprier22a.pdf (623.01 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03736116 , version 1 (22-07-2022)

Licence

Paternité

Identifiants

  • HAL Id : hal-03736116 , version 1

Citer

Sylvain Lamprier, Thomas Scialom, Antoine Chaffin, Vincent Claveau, Ewa Kijak, et al.. Generative Cooperative Networks for Natural Language Generation. ICML 2022 - 39th International Conference on Machine Learning, Jul 2022, Baltimore, MD, United States. pp.11891--11905. ⟨hal-03736116⟩
91 Consultations
113 Téléchargements

Partager

Gmail Facebook X LinkedIn More