W15-4651 investigated the convergence speed of the online adaptation . We conducted the following
P13-1082 translation model framework with the online adaptation of other models , or the log-linear
W13-2237 feature values , we can also apply online adaptation on the feature weights of the
W15-4651 of available utterances for the online adaptation increased . Vertical and horizontal
W95-0108 our experiments . Pruning during online adaptation has two advantages . First ,
W13-2237 2012 ) presented a comparison of online adaptation techniques in post editing scenario
W15-4651 target utterance . We call this " online adaptation " . We also virtually calculated
N07-4014 this offers the potential for online adaptation with real users at a later stage
P13-1046 also hope to explore unsupervised online adaptation , where the trained model can
W99-0627 Lewis and Gale , 1994 ) . Such online adaptation can maximally leverage program
N10-1062 all word alignments . For the online adaptation experiments we modified Model
H91-1088 an interface which allows for online adaptation . This interface constitutes
W13-2237 ) improved SMT performance by online adaptation of scaling factors ( A in ( 1
P14-1083 each interaction is too costly , online adaptation after each interaction has become
W09-1602 Besides , that would also enable online adaptation of the model parameters of the
P14-1067 . For this reason , we use the online adaptation of E-SVR proposed by ( Ma et
W15-4651 classification performance is unstable in online adaptation when the number of available
W15-4651 the closed tests , those of the online adaptation were calculated also as the closed
W95-0108 ) and of being able to perform online adaptation . Moreover , the interpolation
W15-4651 batch adaptation were higher than online adaptation conditions by 3.1 and 2.8 percentage
hide detail