Matches in SemOpenAlex for { <https://semopenalex.org/work/W68992276> ?p ?o ?g. }
- W68992276 abstract "Over the last two decades, large margin methods have yielded excellent performance on many tasks. The theoretical properties of large margin methods have been intensively studied and are especially well-established for support vector machines (SVMs). However, the scalability of large margin methods remains an issue due to the amount of computation they require. This is especially true for applications involving sequential data.In this thesis we are motivated by the problem of automatic speech recognition (ASR) whose large-scale applications involve training and testing on extremely large data sets. The acoustic models used in ASR are based on continuous-density hidden Markov models (CD-HMMs). Researchers in ASR have focused on discriminative training of HMMs, which leads to models with significantly lower error rates. More recently, building on the successes of SVMs and various extensions thereof in the machine learning community, a number of researchers in ASR have also explored large margin methods for discriminative training of HMMs.This dissertation aims to apply various large margin methods developed in the machine learning community to the challenging large-scale problems that arise in ASR. Specifically, we explore the use of sequential, mistake-driven updates for online learning and acoustic feature adaptation in large margin HMMs. The updates are applied to the parameters of acoustic models after the decoding of individual training utterances. For large margin training, the updates attempt to separate the log-likelihoods of correct and incorrect transcriptions by an amount proportional to their Hamming distance. For acoustic feature adaptation, the updates attempt to improve recognition by linearly transforming the features computed by the front end. We evaluate acoustic models trained in this way on the TIMIT speech database. We find that online updates for large margin training not only converge faster than analogous batch optimizations, but also yield lower phone error rates than approaches that do not attempt to enforce a large margin.We conclude this thesis with a discussion of future research directions, highlighting in particular the challenges of scaling our approach to the most difficult problems in large-vocabulary continuous speech recognition." @default.
- W68992276 created "2016-06-24" @default.
- W68992276 creator A5029087418 @default.
- W68992276 creator A5059411202 @default.
- W68992276 date "2011-01-01" @default.
- W68992276 modified "2023-09-23" @default.
- W68992276 title "Online learning of large margin hidden markov models for automatic speech recognition" @default.
- W68992276 cites W112175133 @default.
- W68992276 cites W145744102 @default.
- W68992276 cites W1481751294 @default.
- W68992276 cites W1507239162 @default.
- W68992276 cites W1512098439 @default.
- W68992276 cites W1530699444 @default.
- W68992276 cites W1548139318 @default.
- W68992276 cites W1553271421 @default.
- W68992276 cites W1554663460 @default.
- W68992276 cites W1559312645 @default.
- W68992276 cites W1560013842 @default.
- W68992276 cites W1574862351 @default.
- W68992276 cites W1598851216 @default.
- W68992276 cites W1602492977 @default.
- W68992276 cites W1614659291 @default.
- W68992276 cites W170098597 @default.
- W68992276 cites W1709057194 @default.
- W68992276 cites W1714704734 @default.
- W68992276 cites W1762008180 @default.
- W68992276 cites W1865845602 @default.
- W68992276 cites W187290754 @default.
- W68992276 cites W1877570817 @default.
- W68992276 cites W1902568950 @default.
- W68992276 cites W1936118753 @default.
- W68992276 cites W194033037 @default.
- W68992276 cites W1970500656 @default.
- W68992276 cites W1975588358 @default.
- W68992276 cites W1975953721 @default.
- W68992276 cites W1979711143 @default.
- W68992276 cites W1988995507 @default.
- W68992276 cites W1992314141 @default.
- W68992276 cites W1996430422 @default.
- W68992276 cites W2000626086 @default.
- W68992276 cites W2002342963 @default.
- W68992276 cites W2005754897 @default.
- W68992276 cites W2008652694 @default.
- W68992276 cites W2022768064 @default.
- W68992276 cites W2027915610 @default.
- W68992276 cites W2031044178 @default.
- W68992276 cites W2031248101 @default.
- W68992276 cites W2032210760 @default.
- W68992276 cites W2033565080 @default.
- W68992276 cites W2038050265 @default.
- W68992276 cites W2040870580 @default.
- W68992276 cites W2044145689 @default.
- W68992276 cites W2044223162 @default.
- W68992276 cites W2045956438 @default.
- W68992276 cites W2063541597 @default.
- W68992276 cites W206549024 @default.
- W68992276 cites W2067474491 @default.
- W68992276 cites W207489108 @default.
- W68992276 cites W2077804127 @default.
- W68992276 cites W2080400971 @default.
- W68992276 cites W2083787158 @default.
- W68992276 cites W2084514013 @default.
- W68992276 cites W2087347434 @default.
- W68992276 cites W2090578816 @default.
- W68992276 cites W2090861223 @default.
- W68992276 cites W2091825929 @default.
- W68992276 cites W2098601596 @default.
- W68992276 cites W2099119623 @default.
- W68992276 cites W2100474170 @default.
- W68992276 cites W2101276256 @default.
- W68992276 cites W2102486516 @default.
- W68992276 cites W2103359087 @default.
- W68992276 cites W2104448323 @default.
- W68992276 cites W2104997912 @default.
- W68992276 cites W2105644991 @default.
- W68992276 cites W2106554350 @default.
- W68992276 cites W2111478553 @default.
- W68992276 cites W2111479622 @default.
- W68992276 cites W2113651538 @default.
- W68992276 cites W2115439799 @default.
- W68992276 cites W2116410915 @default.
- W68992276 cites W2119821739 @default.
- W68992276 cites W2121354157 @default.
- W68992276 cites W2121423746 @default.
- W68992276 cites W2121990650 @default.
- W68992276 cites W2122537498 @default.
- W68992276 cites W2124995446 @default.
- W68992276 cites W2125234026 @default.
- W68992276 cites W2125838338 @default.
- W68992276 cites W2125993116 @default.
- W68992276 cites W2126033156 @default.
- W68992276 cites W2126714083 @default.
- W68992276 cites W2130477805 @default.
- W68992276 cites W2130556178 @default.
- W68992276 cites W2133864802 @default.
- W68992276 cites W2135346934 @default.
- W68992276 cites W2136687243 @default.
- W68992276 cites W2137079242 @default.
- W68992276 cites W2139212933 @default.
- W68992276 cites W2139686264 @default.