feature selection 0.003708945
other feature 0.003649257
feature set 0.003637037
different feature 0.003516956
feature space 0.003483378
possible feature 0.003452004
certain feature 0.003391555
candidate feature 0.0033845
feature sets 0.00338445
current feature 0.003384072
single feature 0.003378205
new feature 0.003374857
large feature 0.003373278
feature templates 0.003367162
simple feature 0.003352865
incremental feature 0.003350697
top feature 0.003341564
corresponding feature 0.003327605
ifs feature 0.003324719
feature subset 0.003324114
quality feature 0.003317718
conclusion feature 0.003313452
feature candidates 0.003300434
feature subsets 0.003293232
feature spaces 0.003272859
whole feature 0.003272821
fast feature 0.003272047
feature tem 0.003271998
feature selec 0.003271038
mental feature 0.003269904
feature indi 0.003266851
feature 0.00305587
other features 0.002921107
different features 0.002788806
good features 0.0026828620000000003
candidate features 0.00265635
frequent features 0.002641498
ing features 0.0026383910000000003
link features 0.002616362
quality features 0.002589568
various features 0.002583553
selecting features 0.002573024
dundant features 0.002540451
rest features 0.0025394560000000003
features 0.00232772
model selection 0.002120125
language model 0.001945212
active model 0.0018029779999999998
exponential model 0.0018019339999999998
current model 0.0017952519999999998
new model 0.0017860369999999999
model adjustment 0.001685797
training data 0.001582237
model 0.00146705
selection algorithm 0.001402184
test data 0.001304285
word link 0.0012441779999999999
same time 0.001094167
new algorithm 0.001068096
wsj data 0.001052281
cutoff algorithm 0.001030439
real training 0.001018115
ifs algorithm 0.001017958
test corpus 0.001010204
sgc algorithm 0.001008873
second algorithm 0.001008169
training instance 0.001002765
tion algorithm 0.001000944
training instances 0.00100018
previous models 9.94611E-4
last algorithm 9.94041E-4
selection technique 9.85848E-4
whole training 9.80286E-4
algorithm com 9.73962E-4
selection algorithms 9.65491E-4
fast algorithm 9.65286E-4
refined algorithm 9.605E-4
ture selection 9.59985E-4
entire selection 9.56006E-4
different values 9.49421E-4
selection algo 9.48582E-4
selection processing 9.45768E-4
selection stage 9.442420000000001E-4
selection point 9.418180000000001E-4
parameter values 9.28551E-4
selection process 9.175780000000001E-4
log time 8.91409E-4
whole selection 8.700260000000001E-4
test score 8.62678E-4
empty set 8.587229999999999E-4
language modeling 8.57463E-4
other steps 8.52402E-4
other hand 8.52112E-4
memory size 8.41319E-4
prior distribution 8.35443E-4
same framework 8.25578E-4
instance set 8.20597E-4
parameter estimation 8.15649E-4
same qual 8.04536E-4
same thing 8.04536E-4
