mert algorithm 0.0017588089999999998
search algorithm 0.001698168
such feature 0.001585415
other feature 0.001564303
line search 0.001510379
feature vector 0.0015045599999999998
language data 0.0014737930000000002
function word 0.001469489
training data 0.001370941
exact algorithm 0.001361216
feature functions 0.001330945
simplex algorithm 0.0013288549999999999
efficient algorithm 0.001293084
optimal feature 0.0012735889999999999
och algorithm 0.0012705499999999998
algorithm polynomial 0.0012610009999999999
feature sets 0.001257887
single line 0.001248655
objective function 0.001244612
language model 0.001244483
machine translation 0.0012397900000000002
feature vectors 0.001231806
score function 0.001226935
constant function 0.00121832
line graph 0.001208032
latter function 0.001185044
language models 0.0011811249999999999
word error 0.001172273
translation systems 0.001163521
translation system 0.001159602
search space 0.0011571279999999999
special line 0.001154479
translation performance 0.001154027
parameter vector 0.001148325
test data 0.001140861
function getnextbest 0.0011343059999999999
function expandfrontier 0.00112771
natural line 0.001123827
linear model 0.0011187060000000001
line min 0.0011147659999999999
line segment 0.001103668
first case 0.0010996539999999998
line searches 0.001077418
optimization time 0.001073321
open line 0.001070513
weight vector 0.00106707
mert let 0.001064396
translation pair 0.001060823
other research 0.001058276
translation combinations 0.001056296
standard set 0.001054411
translation hypotheses 0.001043286
algorithm 0.00103467
data sizes 0.001031088
other evaluation 0.001026707
parameter space 0.0010226179999999999
first approach 0.001020265
new set 0.001011633
optimal vector 0.001007703
monolingual data 9.99752E-4
ment data 9.97278E-4
parliamentary data 9.961380000000001E-4
task loss 9.96114E-4
approximate search 9.94495E-4
source language 9.94418E-4
exact search 9.90044E-4
global search 9.88976E-4
language applications 9.84611E-4
optimization problem 9.83264E-4
natural language 9.77133E-4
mert comparison 9.738549999999999E-4
fast mert 9.691629999999999E-4
test set 9.68769E-4
reference translation 9.669990000000001E-4
search directions 9.63551E-4
mert finding 9.618179999999999E-4
such templates 9.59122E-4
objective vector 9.57938E-4
search errors 9.536799999999999E-4
tion set 9.519460000000001E-4
translation mappings 9.5128E-4
german translation 9.5128E-4
machine learning 9.49449E-4
source tree 9.487079999999999E-4
tuning set 9.47407E-4
multidimensional mert 9.46872E-4
unidimensional mert 9.46872E-4
vector combinations 9.46505E-4
such representations 9.40759E-4
search algo 9.352130000000001E-4
general case 9.34343E-4
convex set 9.31541E-4
loss functions 9.21272E-4
similar approach 9.17455E-4
model parameters 9.127639999999999E-4
hidden state 9.11769E-4
test time 9.072119999999999E-4
function 9.06011E-4
previous work 9.05441E-4
search techniques 9.04012E-4
