Matches in SemOpenAlex for { <https://semopenalex.org/work/W3144486013> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W3144486013 abstract "In machine learning, training multiple models for the same task, and using the outputs from all the models helps reduce the variance of the combined result. Using an ensemble of models in classification tasks such as Automatic Speech Recognition (ASR) improves the accuracy across different target domains such as multiple accents, environmental conditions, and other scenarios. It is possible to select model weights for the ensemble in numerous ways. A classifier trained to identify target domain, a simple averaging function, or an exhaustive grid search are the common approaches to obtain suitable weights. All these methods suffer either in choosing sub-optimal weights or by being computationally expensive. We propose a novel and practical method for dynamic weight selection in an ensemble, which can approximate a grid search in a time-efficient manner. We show that a combination of weights always performs better than assigning uniform weights for all models. Our algorithm can utilize a validation set if available or find weights dynamically from the input utterance itself. Experiments conducted for various ASR tasks show that the proposed method outperforms the uniformly weighted ensemble in terms of Word Error Rate (WER) in our experiments." @default.
- W3144486013 created "2021-04-13" @default.
- W3144486013 creator A5000757079 @default.
- W3144486013 creator A5011249075 @default.
- W3144486013 creator A5021662681 @default.
- W3144486013 creator A5024123814 @default.
- W3144486013 creator A5048344182 @default.
- W3144486013 date "2021-01-19" @default.
- W3144486013 modified "2023-10-13" @default.
- W3144486013 title "Dynamically Weighted Ensemble Models for Automatic Speech Recognition" @default.
- W3144486013 cites W1494198834 @default.
- W3144486013 cites W2005708641 @default.
- W3144486013 cites W2062164080 @default.
- W3144486013 cites W2064675550 @default.
- W3144486013 cites W2134006700 @default.
- W3144486013 cites W2150884987 @default.
- W3144486013 cites W2160815625 @default.
- W3144486013 cites W2508418541 @default.
- W3144486013 cites W2616139854 @default.
- W3144486013 cites W2972539100 @default.
- W3144486013 cites W4212883601 @default.
- W3144486013 doi "https://doi.org/10.1109/slt48900.2021.9383463" @default.
- W3144486013 hasPublicationYear "2021" @default.
- W3144486013 type Work @default.
- W3144486013 sameAs 3144486013 @default.
- W3144486013 citedByCount "1" @default.
- W3144486013 countsByYear W31444860132023 @default.
- W3144486013 crossrefType "proceedings-article" @default.
- W3144486013 hasAuthorship W3144486013A5000757079 @default.
- W3144486013 hasAuthorship W3144486013A5011249075 @default.
- W3144486013 hasAuthorship W3144486013A5021662681 @default.
- W3144486013 hasAuthorship W3144486013A5024123814 @default.
- W3144486013 hasAuthorship W3144486013A5048344182 @default.
- W3144486013 hasConcept C10485038 @default.
- W3144486013 hasConcept C119857082 @default.
- W3144486013 hasConcept C119898033 @default.
- W3144486013 hasConcept C121955636 @default.
- W3144486013 hasConcept C12267149 @default.
- W3144486013 hasConcept C144133560 @default.
- W3144486013 hasConcept C153180895 @default.
- W3144486013 hasConcept C154945302 @default.
- W3144486013 hasConcept C162324750 @default.
- W3144486013 hasConcept C177264268 @default.
- W3144486013 hasConcept C187691185 @default.
- W3144486013 hasConcept C187736073 @default.
- W3144486013 hasConcept C196083921 @default.
- W3144486013 hasConcept C199360897 @default.
- W3144486013 hasConcept C2524010 @default.
- W3144486013 hasConcept C2780451532 @default.
- W3144486013 hasConcept C33923547 @default.
- W3144486013 hasConcept C40969351 @default.
- W3144486013 hasConcept C41008148 @default.
- W3144486013 hasConcept C45942800 @default.
- W3144486013 hasConcept C95623464 @default.
- W3144486013 hasConceptScore W3144486013C10485038 @default.
- W3144486013 hasConceptScore W3144486013C119857082 @default.
- W3144486013 hasConceptScore W3144486013C119898033 @default.
- W3144486013 hasConceptScore W3144486013C121955636 @default.
- W3144486013 hasConceptScore W3144486013C12267149 @default.
- W3144486013 hasConceptScore W3144486013C144133560 @default.
- W3144486013 hasConceptScore W3144486013C153180895 @default.
- W3144486013 hasConceptScore W3144486013C154945302 @default.
- W3144486013 hasConceptScore W3144486013C162324750 @default.
- W3144486013 hasConceptScore W3144486013C177264268 @default.
- W3144486013 hasConceptScore W3144486013C187691185 @default.
- W3144486013 hasConceptScore W3144486013C187736073 @default.
- W3144486013 hasConceptScore W3144486013C196083921 @default.
- W3144486013 hasConceptScore W3144486013C199360897 @default.
- W3144486013 hasConceptScore W3144486013C2524010 @default.
- W3144486013 hasConceptScore W3144486013C2780451532 @default.
- W3144486013 hasConceptScore W3144486013C33923547 @default.
- W3144486013 hasConceptScore W3144486013C40969351 @default.
- W3144486013 hasConceptScore W3144486013C41008148 @default.
- W3144486013 hasConceptScore W3144486013C45942800 @default.
- W3144486013 hasConceptScore W3144486013C95623464 @default.
- W3144486013 hasLocation W31444860131 @default.
- W3144486013 hasOpenAccess W3144486013 @default.
- W3144486013 hasPrimaryLocation W31444860131 @default.
- W3144486013 hasRelatedWork W1807784185 @default.
- W3144486013 hasRelatedWork W1909207154 @default.
- W3144486013 hasRelatedWork W2794896638 @default.
- W3144486013 hasRelatedWork W3101614107 @default.
- W3144486013 hasRelatedWork W3124390867 @default.
- W3144486013 hasRelatedWork W3149839747 @default.
- W3144486013 hasRelatedWork W3202800081 @default.
- W3144486013 hasRelatedWork W3204228978 @default.
- W3144486013 hasRelatedWork W3210229324 @default.
- W3144486013 hasRelatedWork W45170056 @default.
- W3144486013 isParatext "false" @default.
- W3144486013 isRetracted "false" @default.
- W3144486013 magId "3144486013" @default.
- W3144486013 workType "article" @default.