D08-1082 best candidate predictions with a discriminative reranking algorithm . m ∗ = arg max
D13-1111 Reranking Experiments We now turn to discriminative reranking , which has frequently been used
D11-1102 We use these hypotheses in the discriminative reranking model , instead of the original
D10-1003 extra-sentential context into a discriminative reranking parser , which naturally allows
D09-1043 Abstract This paper shows that discriminative reranking with an averaged perceptron model
D08-1082 , 2002 ; Collins , 2001 ) for discriminative reranking . The detailed algorithm can
D12-1040 want to investigate the use of discriminative reranking ( Collins , 2000 ) , which has
D09-1043 this paper , we have shown how discriminative reranking with an averaged perceptron model
D10-1002 WSJ test set and surpass even discriminative reranking systems without self - training
D08-1095 and Koo , 2005 ) . In essence , discriminative reranking allows the re-ordering of results
D09-1112 Conclusions In this paper , we propose discriminative reranking of concept annotation to jointly
D14-1076 skip-chain CRF ( Galley , 2006 ) , discriminative reranking ( Aker et al. , 2010 ) , among
D13-1047 fields ( CRF ) ( Galley , 2006 ) , discriminative reranking ( Aker et al. , 2010 ) , among
D08-1082 the model , when coupled with a discriminative reranking tech - nique , achieves state-of-the-art
D08-1082 this approach , augmented with a discriminative reranking technique , achieves state-of-the-art
D10-1002 work . but again without using a discriminative reranking step . We expect that replacing
D09-1087 combining the PCFG-LA parser with discriminative reranking approaches ( Charniak and Johnson
D11-1102 task . Section 3 describes our discriminative reranking framework for SLU , in particular
D11-1148 introduce in this paper , of using discriminative reranking features as a broader characterisation
D10-1002 competitively to the self-trained two-step discriminative reranking parser of McClosky et al. ( 2006
hide detail