Multi-way Variational NMT for UGC: Improving Robustness in Zero-shot Scenarios via Mixture Density Networks - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2023

Multi-way Variational NMT for UGC: Improving Robustness in Zero-shot Scenarios via Mixture Density Networks

Abstract

This work presents a novel Variational Neural Machine Translation (VNMT) architecture with enhanced robustness properties, which we investigate through a detailed case-study addressing noisy French user-generated content (UGC) translation to English. We show that the proposed model, with results comparable or superior to state-of-the-art VNMT, improves performance over UGC translation in a zero-shot evaluation scenario while keeping optimal translation scores on in-domain test sets. We elaborate on such results by visualizing and explaining how neural learning representations behave when processing UGC noise. In addition, we show that VNMT enforces robustness to the learned embeddings, which can be later used for robust transfer learning approaches.
Fichier principal
Vignette du fichier
_Article__VNMT_nodalida.pdf (1.36 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04384748 , version 1 (10-01-2024)

Identifiers

  • HAL Id : hal-04384748 , version 1

Cite

José Carlos Rosales Núñez, Djamé Seddah, Guillaume Wisniewski. Multi-way Variational NMT for UGC: Improving Robustness in Zero-shot Scenarios via Mixture Density Networks. NoDaLiDa 2023 - 24th Nordic Conference on Computational Linguistics, May 2023, Torshavn, Faroe Islands. ⟨hal-04384748⟩
30 View
10 Download

Share

Gmail Facebook X LinkedIn More