Matches in SemOpenAlex for { <https://semopenalex.org/work/W2050971845> ?p ?o ?g. }
Showing items 1 to 82 of
82
with 100 items per page.
- W2050971845 abstract "In (Chen, 2009), we show that for a variety of language models belonging to the exponential family, the test set cross-entropy of a model can be accurately predicted from its training set cross-entropy and its parameter values. In this work, we show how this relationship can be used to motivate two heuristics for shrinking the size of a language model to improve its performance. We use the first heuristic to develop a novel class-based language model that outperforms a baseline word trigram model by 28% in perplexity and 1.9% absolute in speech recognition word-error rate on Wall Street Journal data. We use the second heuristic to motivate a regularized version of minimum discrimination information models and show that this method outperforms other techniques for domain adaptation." @default.
- W2050971845 created "2016-06-24" @default.
- W2050971845 creator A5008850932 @default.
- W2050971845 date "2009-01-01" @default.
- W2050971845 modified "2023-09-23" @default.
- W2050971845 title "Shrinking exponential language models" @default.
- W2050971845 cites W140816929 @default.
- W2050971845 cites W1481113913 @default.
- W2050971845 cites W169286004 @default.
- W2050971845 cites W1718044877 @default.
- W2050971845 cites W1797288984 @default.
- W2050971845 cites W1823613926 @default.
- W2050971845 cites W182840523 @default.
- W2050971845 cites W1974515274 @default.
- W2050971845 cites W1991696363 @default.
- W2050971845 cites W2020073413 @default.
- W2050971845 cites W2024490156 @default.
- W2050971845 cites W2049901611 @default.
- W2050971845 cites W2069699492 @default.
- W2050971845 cites W2072223048 @default.
- W2050971845 cites W2075401516 @default.
- W2050971845 cites W2076467305 @default.
- W2050971845 cites W2097003869 @default.
- W2050971845 cites W2111305191 @default.
- W2050971845 cites W2117793990 @default.
- W2050971845 cites W2121227244 @default.
- W2050971845 cites W2127095586 @default.
- W2050971845 cites W2133486044 @default.
- W2050971845 cites W2134237567 @default.
- W2050971845 cites W2143866356 @default.
- W2050971845 cites W2145826596 @default.
- W2050971845 cites W2155388323 @default.
- W2050971845 cites W2158195707 @default.
- W2050971845 cites W2489195900 @default.
- W2050971845 cites W98731357 @default.
- W2050971845 doi "https://doi.org/10.3115/1620754.1620822" @default.
- W2050971845 hasPublicationYear "2009" @default.
- W2050971845 type Work @default.
- W2050971845 sameAs 2050971845 @default.
- W2050971845 citedByCount "88" @default.
- W2050971845 countsByYear W20509718452012 @default.
- W2050971845 countsByYear W20509718452013 @default.
- W2050971845 countsByYear W20509718452014 @default.
- W2050971845 countsByYear W20509718452015 @default.
- W2050971845 countsByYear W20509718452016 @default.
- W2050971845 countsByYear W20509718452017 @default.
- W2050971845 countsByYear W20509718452018 @default.
- W2050971845 countsByYear W20509718452019 @default.
- W2050971845 countsByYear W20509718452020 @default.
- W2050971845 countsByYear W20509718452021 @default.
- W2050971845 countsByYear W20509718452023 @default.
- W2050971845 crossrefType "proceedings-article" @default.
- W2050971845 hasAuthorship W2050971845A5008850932 @default.
- W2050971845 hasBestOaLocation W20509718451 @default.
- W2050971845 hasConcept C134306372 @default.
- W2050971845 hasConcept C151376022 @default.
- W2050971845 hasConcept C28826006 @default.
- W2050971845 hasConcept C33923547 @default.
- W2050971845 hasConcept C41008148 @default.
- W2050971845 hasConceptScore W2050971845C134306372 @default.
- W2050971845 hasConceptScore W2050971845C151376022 @default.
- W2050971845 hasConceptScore W2050971845C28826006 @default.
- W2050971845 hasConceptScore W2050971845C33923547 @default.
- W2050971845 hasConceptScore W2050971845C41008148 @default.
- W2050971845 hasLocation W20509718451 @default.
- W2050971845 hasLocation W20509718452 @default.
- W2050971845 hasOpenAccess W2050971845 @default.
- W2050971845 hasPrimaryLocation W20509718451 @default.
- W2050971845 hasRelatedWork W1485733024 @default.
- W2050971845 hasRelatedWork W1655909868 @default.
- W2050971845 hasRelatedWork W1973610729 @default.
- W2050971845 hasRelatedWork W2051922179 @default.
- W2050971845 hasRelatedWork W2332276036 @default.
- W2050971845 hasRelatedWork W2369988800 @default.
- W2050971845 hasRelatedWork W2387796032 @default.
- W2050971845 hasRelatedWork W2793712274 @default.
- W2050971845 hasRelatedWork W2999388306 @default.
- W2050971845 hasRelatedWork W4287905284 @default.
- W2050971845 isParatext "false" @default.
- W2050971845 isRetracted "false" @default.
- W2050971845 magId "2050971845" @default.
- W2050971845 workType "article" @default.