D10-1095 section , we used the confidence estimation algorithms to choose individual words to
H01-1058 Besides data sparseness , the estimation algorithms ( e.g. EM algorithm ) might be
C00-1056 called , Forward-Backward parameter estimation algorithm , is used by Sharman in phonetic
D14-1210 statistically significant . The two estimation algorithms differ significantly in their
C04-1041 parallel implementation of the estimation algorithm which runs on a Beowulf cluster
D09-1074 The corpus weight approach and estimation algorithm are described in Section 3 .
D15-1214 Conclusion We presented a novel estimation algorithm for latent-variable PCFGs . This
D11-1137 forest . Algorithm 2 shows the estimation algorithm . First , it runs the inside/outside
D15-1214 perturbing the underlying features that estimation algorithm uses from the training data .
D15-1104 McCallum , 2006 ) The parameter estimation algorithm is abstracted in Algorithm 1
C00-2121 word . The context similarity estimation algorithms were run using vocabularies of
D12-1125 in a simple , EM-like parameter estimation algorithm as discussed in Section 5 . To
E14-1037 UBL 's gradient-based parameter estimation algorithm tries to maximize ET P ( L ,
D11-1005 expectations can readily be used in estimation algorithms such as expectationmaximization
E03-1071 further the already very simple estimation algorithm . Although GIS is not as fast
E12-1024 use of an incremental parameter estimation algorithm , online Variational Bayesian
A00-2020 of the sparse Markov transducer estimation algorithm is to estimate a conditional
D15-1214 used in the past for other L-PCFG estimation algorithms . Cohen et al. ( 2012 ) used
D14-1210 likelihood criterion . The parameter estimation algorithms are relatively similar , but
C94-2210 the Forward - Backward parameter estimation algorithm \ -LSB- 7 \ -RSB- can be used
hide detail