sequence features 0.0030612689999999997
several features 0.003005721
work features 0.0029674099999999998
rule features 0.0029489729999999997
modeling features 0.002944005
additional features 0.00293052
language model 0.0029074120000000003
syntactic features 0.002895243
ing features 0.002869562
simple features 0.002842817
features condition 0.002799129
example features 0.002784577
gram features 0.002738744
complicated features 0.0027329159999999997
arbitrary features 0.00272612
dicator features 0.002705748
positional features 0.002705748
selecting features 0.002705748
increments features 0.002705748
possible feature 0.002619818
reranking model 0.002599745
feature vector 0.002545127
new model 0.0025342330000000003
tagging model 0.002527285
feature selection 0.002509215
model score 0.002507263
acoustic model 0.0024986070000000003
generative model 0.0024929780000000003
model scores 0.002481687
features 0.00247367
feature space 0.002461875
tic model 0.002437713
effective feature 0.002434724
feature examples 0.002391772
feature sets 0.002390128
feature definitions 0.0023671589999999998
feature extraction 0.002360226
tential feature 0.002331701
feature extrac 0.002330366
model 0.0021339
feature 0.00209801
word error 0.001800646
training data 0.001716464
initial word 0.0017026440000000001
training set 0.001667066
training algorithm 0.001650011
perceptron training 0.001642367
next word 0.001605128
training method 0.001512822
discriminative training 0.001464478
parameter estimation 0.001461098
language models 0.001400937
perceptron results 0.001390656
asr training 0.001368052
large parameter 0.00134236
pos perceptron 0.0013389209999999999
entire training 0.001322234
training example 0.001318497
training algorithms 0.001298015
first parameter 0.001292362
perceptron algorithm 0.001277198
training conditions 0.001276575
discriminative parameter 0.0012653669999999999
previous words 0.0012634550000000001
training utterances 0.0012567160000000002
parameter vector 0.0012555959999999999
training lattices 0.001253323
pos tag 0.0012504489999999998
training utterance 0.001249997
second parameter 0.001249847
ceptron training 0.001245915
language modeling 0.001243847
training exam 0.001240292
fast training 0.001239693
discriminative language 0.0012304
test data 0.00122087
syntactic language 0.001195085
linear models 0.001181165
test set 0.001171472
parameter values 0.001162954
first approach 0.001159946
estimation method 0.001157851
syntactic information 0.001151891
empirical results 0.001145458
reranking approach 0.001141908
speech recognition 0.001135024
sequence tag 0.0011339039999999998
tag sequence 0.0011339039999999998
baseline speech 0.001131068
alternative parameter 0.001130257
second set 0.001100844
pos tags 0.00109964
parse sequence 0.001082745
first sequence 0.001071482
switchboard speech 0.0010688289999999999
trigram language 0.001063847
gclm results 0.001056326
ing set 0.001055368
estimation methods 0.001054749
conditional probability 0.001046768
