Matches in SemOpenAlex for { <https://semopenalex.org/work/W1502105280> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W1502105280 abstract "Language models (LMs) are often constructed by building multiple individual component models that are combined using context independent interpolation weights. By tuning these weights, using either perplexity or discriminative approaches, it is possible to adapt LMs to a particular task. This paper investigates the use of context dependent weighting in both interpolation and test-time adaptation of language models. Depending on the previous word contexts, a discrete history weighting function is used to adjust the contribution from each component model. As this dramatically increases the number of parameters to estimate, robust weight estimation schemes are required. Several approaches are described in this paper. The first approach is based on MAP estimation where interpolation weights of lower order contexts are used as smoothing priors. The second approach uses training data to ensure robust estimation of LM interpolation weights. This can also serve as a smoothing prior for MAP adaptation. A normalized perplexity metric is proposed to handle the bias of the standard perplexity criterion to corpus size. A range of schemes to combine weight information obtained from training data and test data hypotheses are also proposed to improve robustness during context dependent LM adaptation. In addition, a minimum Bayes' risk (MBR) based discriminative training scheme is also proposed. An efficient weighted finite state transducer (WFST) decoding algorithm for context dependent interpolation is also presented. The proposed technique was evaluated using a state-of-the-art Mandarin Chinese broadcast speech transcription task. Character error rate (CER) reductions up to 7.3% relative were obtained as well as consistent perplexity improvements." @default.
- W1502105280 created "2016-06-24" @default.
- W1502105280 creator A5002191410 @default.
- W1502105280 creator A5037109470 @default.
- W1502105280 creator A5050766679 @default.
- W1502105280 date "2009-09-06" @default.
- W1502105280 modified "2023-09-24" @default.
- W1502105280 title "Use of contexts in language model interpolation and adaptation" @default.
- W1502105280 doi "https://doi.org/10.21437/interspeech.2009-115" @default.
- W1502105280 hasPublicationYear "2009" @default.
- W1502105280 type Work @default.
- W1502105280 sameAs 1502105280 @default.
- W1502105280 citedByCount "7" @default.
- W1502105280 countsByYear W15021052802012 @default.
- W1502105280 countsByYear W15021052802017 @default.
- W1502105280 countsByYear W15021052802018 @default.
- W1502105280 crossrefType "proceedings-article" @default.
- W1502105280 hasAuthorship W1502105280A5002191410 @default.
- W1502105280 hasAuthorship W1502105280A5037109470 @default.
- W1502105280 hasAuthorship W1502105280A5050766679 @default.
- W1502105280 hasConcept C100279451 @default.
- W1502105280 hasConcept C104114177 @default.
- W1502105280 hasConcept C104317684 @default.
- W1502105280 hasConcept C11413529 @default.
- W1502105280 hasConcept C119857082 @default.
- W1502105280 hasConcept C126838900 @default.
- W1502105280 hasConcept C137293760 @default.
- W1502105280 hasConcept C137800194 @default.
- W1502105280 hasConcept C154945302 @default.
- W1502105280 hasConcept C183115368 @default.
- W1502105280 hasConcept C185592680 @default.
- W1502105280 hasConcept C28490314 @default.
- W1502105280 hasConcept C31972630 @default.
- W1502105280 hasConcept C3770464 @default.
- W1502105280 hasConcept C41008148 @default.
- W1502105280 hasConcept C55493867 @default.
- W1502105280 hasConcept C63479239 @default.
- W1502105280 hasConcept C71924100 @default.
- W1502105280 hasConcept C97931131 @default.
- W1502105280 hasConceptScore W1502105280C100279451 @default.
- W1502105280 hasConceptScore W1502105280C104114177 @default.
- W1502105280 hasConceptScore W1502105280C104317684 @default.
- W1502105280 hasConceptScore W1502105280C11413529 @default.
- W1502105280 hasConceptScore W1502105280C119857082 @default.
- W1502105280 hasConceptScore W1502105280C126838900 @default.
- W1502105280 hasConceptScore W1502105280C137293760 @default.
- W1502105280 hasConceptScore W1502105280C137800194 @default.
- W1502105280 hasConceptScore W1502105280C154945302 @default.
- W1502105280 hasConceptScore W1502105280C183115368 @default.
- W1502105280 hasConceptScore W1502105280C185592680 @default.
- W1502105280 hasConceptScore W1502105280C28490314 @default.
- W1502105280 hasConceptScore W1502105280C31972630 @default.
- W1502105280 hasConceptScore W1502105280C3770464 @default.
- W1502105280 hasConceptScore W1502105280C41008148 @default.
- W1502105280 hasConceptScore W1502105280C55493867 @default.
- W1502105280 hasConceptScore W1502105280C63479239 @default.
- W1502105280 hasConceptScore W1502105280C71924100 @default.
- W1502105280 hasConceptScore W1502105280C97931131 @default.
- W1502105280 hasLocation W15021052801 @default.
- W1502105280 hasOpenAccess W1502105280 @default.
- W1502105280 hasPrimaryLocation W15021052801 @default.
- W1502105280 hasRelatedWork W117095711 @default.
- W1502105280 hasRelatedWork W119114634 @default.
- W1502105280 hasRelatedWork W170098597 @default.
- W1502105280 hasRelatedWork W2127049133 @default.
- W1502105280 hasRelatedWork W217970951 @default.
- W1502105280 hasRelatedWork W587794757 @default.
- W1502105280 hasRelatedWork W82936479 @default.
- W1502105280 hasRelatedWork W1034274924 @default.
- W1502105280 hasRelatedWork W1483599758 @default.
- W1502105280 hasRelatedWork W1486520967 @default.
- W1502105280 hasRelatedWork W1489249682 @default.
- W1502105280 hasRelatedWork W1501432911 @default.
- W1502105280 hasRelatedWork W1508471544 @default.
- W1502105280 hasRelatedWork W1510021464 @default.
- W1502105280 hasRelatedWork W1514870515 @default.
- W1502105280 hasRelatedWork W1515197513 @default.
- W1502105280 hasRelatedWork W275939094 @default.
- W1502105280 hasRelatedWork W299739017 @default.
- W1502105280 hasRelatedWork W951908411 @default.
- W1502105280 hasRelatedWork W964006179 @default.
- W1502105280 isParatext "false" @default.
- W1502105280 isRetracted "false" @default.
- W1502105280 magId "1502105280" @default.
- W1502105280 workType "article" @default.