Matches in SemOpenAlex for { <https://semopenalex.org/work/W4306802327> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W4306802327 abstract "k-Nearest-Neighbor Machine Translation (kNN-MT) becomes an important research direction of NMT in recent years. Its main idea is to retrieve useful key-value pairs from an additional datastore to modify translations without updating the NMT model. However, the underlying retrieved noisy pairs will dramatically deteriorate the model performance. In this paper, we conduct a preliminary study and find that this problem results from not fully exploiting the prediction of the NMT model. To alleviate the impact of noise, we propose a confidence-enhanced kNN-MT model with robust training. Concretely, we introduce the NMT confidence to refine the modeling of two important components of kNN-MT: kNN distribution and the interpolation weight. Meanwhile we inject two types of perturbations into the retrieved pairs for robust training. Experimental results on four benchmark datasets demonstrate that our model not only achieves significant improvements over current kNN-MT models, but also exhibits better robustness. Our code is available at https://github.com/DeepLearnXMU/Robust-knn-mt." @default.
- W4306802327 created "2022-10-20" @default.
- W4306802327 creator A5010357390 @default.
- W4306802327 creator A5024849044 @default.
- W4306802327 creator A5046871248 @default.
- W4306802327 creator A5056683457 @default.
- W4306802327 creator A5066326238 @default.
- W4306802327 creator A5070134243 @default.
- W4306802327 creator A5083081872 @default.
- W4306802327 date "2022-10-17" @default.
- W4306802327 modified "2023-10-16" @default.
- W4306802327 title "Towards Robust k-Nearest-Neighbor Machine Translation" @default.
- W4306802327 doi "https://doi.org/10.48550/arxiv.2210.08808" @default.
- W4306802327 hasPublicationYear "2022" @default.
- W4306802327 type Work @default.
- W4306802327 citedByCount "0" @default.
- W4306802327 crossrefType "posted-content" @default.
- W4306802327 hasAuthorship W4306802327A5010357390 @default.
- W4306802327 hasAuthorship W4306802327A5024849044 @default.
- W4306802327 hasAuthorship W4306802327A5046871248 @default.
- W4306802327 hasAuthorship W4306802327A5056683457 @default.
- W4306802327 hasAuthorship W4306802327A5066326238 @default.
- W4306802327 hasAuthorship W4306802327A5070134243 @default.
- W4306802327 hasAuthorship W4306802327A5083081872 @default.
- W4306802327 hasBestOaLocation W43068023271 @default.
- W4306802327 hasConcept C104317684 @default.
- W4306802327 hasConcept C105580179 @default.
- W4306802327 hasConcept C113238511 @default.
- W4306802327 hasConcept C115961682 @default.
- W4306802327 hasConcept C119857082 @default.
- W4306802327 hasConcept C124101348 @default.
- W4306802327 hasConcept C13280743 @default.
- W4306802327 hasConcept C149364088 @default.
- W4306802327 hasConcept C153180895 @default.
- W4306802327 hasConcept C154945302 @default.
- W4306802327 hasConcept C177264268 @default.
- W4306802327 hasConcept C185592680 @default.
- W4306802327 hasConcept C185798385 @default.
- W4306802327 hasConcept C199360897 @default.
- W4306802327 hasConcept C203005215 @default.
- W4306802327 hasConcept C205649164 @default.
- W4306802327 hasConcept C26517878 @default.
- W4306802327 hasConcept C2776760102 @default.
- W4306802327 hasConcept C2781170535 @default.
- W4306802327 hasConcept C38652104 @default.
- W4306802327 hasConcept C41008148 @default.
- W4306802327 hasConcept C55493867 @default.
- W4306802327 hasConcept C63479239 @default.
- W4306802327 hasConcept C99498987 @default.
- W4306802327 hasConceptScore W4306802327C104317684 @default.
- W4306802327 hasConceptScore W4306802327C105580179 @default.
- W4306802327 hasConceptScore W4306802327C113238511 @default.
- W4306802327 hasConceptScore W4306802327C115961682 @default.
- W4306802327 hasConceptScore W4306802327C119857082 @default.
- W4306802327 hasConceptScore W4306802327C124101348 @default.
- W4306802327 hasConceptScore W4306802327C13280743 @default.
- W4306802327 hasConceptScore W4306802327C149364088 @default.
- W4306802327 hasConceptScore W4306802327C153180895 @default.
- W4306802327 hasConceptScore W4306802327C154945302 @default.
- W4306802327 hasConceptScore W4306802327C177264268 @default.
- W4306802327 hasConceptScore W4306802327C185592680 @default.
- W4306802327 hasConceptScore W4306802327C185798385 @default.
- W4306802327 hasConceptScore W4306802327C199360897 @default.
- W4306802327 hasConceptScore W4306802327C203005215 @default.
- W4306802327 hasConceptScore W4306802327C205649164 @default.
- W4306802327 hasConceptScore W4306802327C26517878 @default.
- W4306802327 hasConceptScore W4306802327C2776760102 @default.
- W4306802327 hasConceptScore W4306802327C2781170535 @default.
- W4306802327 hasConceptScore W4306802327C38652104 @default.
- W4306802327 hasConceptScore W4306802327C41008148 @default.
- W4306802327 hasConceptScore W4306802327C55493867 @default.
- W4306802327 hasConceptScore W4306802327C63479239 @default.
- W4306802327 hasConceptScore W4306802327C99498987 @default.
- W4306802327 hasLocation W43068023271 @default.
- W4306802327 hasOpenAccess W4306802327 @default.
- W4306802327 hasPrimaryLocation W43068023271 @default.
- W4306802327 hasRelatedWork W2146076056 @default.
- W4306802327 hasRelatedWork W2167440101 @default.
- W4306802327 hasRelatedWork W2313858059 @default.
- W4306802327 hasRelatedWork W2883887418 @default.
- W4306802327 hasRelatedWork W2986712369 @default.
- W4306802327 hasRelatedWork W3108681154 @default.
- W4306802327 hasRelatedWork W3115229845 @default.
- W4306802327 hasRelatedWork W3116896278 @default.
- W4306802327 hasRelatedWork W3205677146 @default.
- W4306802327 hasRelatedWork W4220853141 @default.
- W4306802327 isParatext "false" @default.
- W4306802327 isRetracted "false" @default.
- W4306802327 workType "article" @default.