Matches in SemOpenAlex for { <https://semopenalex.org/work/W2892124901> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W2892124901 abstract "The standard evaluation metric of automatic speech recognition (ASR) is the word error rate (WER), which measures the dissimilarity between recognized word sequences and their ground truth. Many training algorithms designed to reduce sequence-level errors such as WER have been proposed for hidden Markov model (HMM)-based ASR, e.g., state-level minimum Bayes risk (sMBR). However, these approaches cannot be used directly for encoder-decoder model based end-to-end ASR, because the encoder-decoder model employs very different mechanisms from HMM-based approaches. In this paper, we propose a new method for optimizing the encoder-decoder model based on a sequence-level evaluation metric. Since the WER is not directly differentiable, we adopt a policy gradient objective function to train the encoder-decoder model, which enables us to minimize the expected WER of the model predictions. This training method employs the scoring of multiple hypotheses as in the decoding stage while usual cross entropy training uses only the ground truth. Therefore, we can expect it to improve the decoding results of the encoder-decoder model. We perform experiments using the Tedlium corpus to demonstrate the potential of our proposed method for improving the recognition performance of the encoder-decoder model." @default.
- W2892124901 created "2018-09-27" @default.
- W2892124901 creator A5021240106 @default.
- W2892124901 creator A5023868166 @default.
- W2892124901 creator A5036532214 @default.
- W2892124901 creator A5080536838 @default.
- W2892124901 date "2018-04-01" @default.
- W2892124901 modified "2023-09-27" @default.
- W2892124901 title "SEQUENCE TRAINING OF ENCODER-DECODER MODEL USING POLICY GRADIENT FOR END- TO-END SPEECH RECOGNITION" @default.
- W2892124901 cites W2143612262 @default.
- W2892124901 cites W2257979135 @default.
- W2892124901 doi "https://doi.org/10.1109/icassp.2018.8462245" @default.
- W2892124901 hasPublicationYear "2018" @default.
- W2892124901 type Work @default.
- W2892124901 sameAs 2892124901 @default.
- W2892124901 citedByCount "14" @default.
- W2892124901 countsByYear W28921249012018 @default.
- W2892124901 countsByYear W28921249012019 @default.
- W2892124901 countsByYear W28921249012020 @default.
- W2892124901 countsByYear W28921249012021 @default.
- W2892124901 countsByYear W28921249012022 @default.
- W2892124901 countsByYear W28921249012023 @default.
- W2892124901 crossrefType "proceedings-article" @default.
- W2892124901 hasAuthorship W2892124901A5021240106 @default.
- W2892124901 hasAuthorship W2892124901A5023868166 @default.
- W2892124901 hasAuthorship W2892124901A5036532214 @default.
- W2892124901 hasAuthorship W2892124901A5080536838 @default.
- W2892124901 hasConcept C111919701 @default.
- W2892124901 hasConcept C11413529 @default.
- W2892124901 hasConcept C118505674 @default.
- W2892124901 hasConcept C137293760 @default.
- W2892124901 hasConcept C146849305 @default.
- W2892124901 hasConcept C153180895 @default.
- W2892124901 hasConcept C154945302 @default.
- W2892124901 hasConcept C162324750 @default.
- W2892124901 hasConcept C167981619 @default.
- W2892124901 hasConcept C176217482 @default.
- W2892124901 hasConcept C185588885 @default.
- W2892124901 hasConcept C21547014 @default.
- W2892124901 hasConcept C23224414 @default.
- W2892124901 hasConcept C28490314 @default.
- W2892124901 hasConcept C40969351 @default.
- W2892124901 hasConcept C41008148 @default.
- W2892124901 hasConcept C57273362 @default.
- W2892124901 hasConcept C74296488 @default.
- W2892124901 hasConceptScore W2892124901C111919701 @default.
- W2892124901 hasConceptScore W2892124901C11413529 @default.
- W2892124901 hasConceptScore W2892124901C118505674 @default.
- W2892124901 hasConceptScore W2892124901C137293760 @default.
- W2892124901 hasConceptScore W2892124901C146849305 @default.
- W2892124901 hasConceptScore W2892124901C153180895 @default.
- W2892124901 hasConceptScore W2892124901C154945302 @default.
- W2892124901 hasConceptScore W2892124901C162324750 @default.
- W2892124901 hasConceptScore W2892124901C167981619 @default.
- W2892124901 hasConceptScore W2892124901C176217482 @default.
- W2892124901 hasConceptScore W2892124901C185588885 @default.
- W2892124901 hasConceptScore W2892124901C21547014 @default.
- W2892124901 hasConceptScore W2892124901C23224414 @default.
- W2892124901 hasConceptScore W2892124901C28490314 @default.
- W2892124901 hasConceptScore W2892124901C40969351 @default.
- W2892124901 hasConceptScore W2892124901C41008148 @default.
- W2892124901 hasConceptScore W2892124901C57273362 @default.
- W2892124901 hasConceptScore W2892124901C74296488 @default.
- W2892124901 hasLocation W28921249011 @default.
- W2892124901 hasOpenAccess W2892124901 @default.
- W2892124901 hasPrimaryLocation W28921249011 @default.
- W2892124901 hasRelatedWork W1966737826 @default.
- W2892124901 hasRelatedWork W2001732961 @default.
- W2892124901 hasRelatedWork W2026408911 @default.
- W2892124901 hasRelatedWork W2143620265 @default.
- W2892124901 hasRelatedWork W2217717732 @default.
- W2892124901 hasRelatedWork W2917344756 @default.
- W2892124901 hasRelatedWork W3011988934 @default.
- W2892124901 hasRelatedWork W4205868073 @default.
- W2892124901 hasRelatedWork W4206374977 @default.
- W2892124901 hasRelatedWork W62743518 @default.
- W2892124901 isParatext "false" @default.
- W2892124901 isRetracted "false" @default.
- W2892124901 magId "2892124901" @default.
- W2892124901 workType "article" @default.