training data 0.00282171
training algorithm 0.002651079
many features 0.002616809
features time 0.002570161
same feature 0.0025121460000000003
active features 0.002438936
training algorithms 0.002265735
feature space 0.002226466
training sample 0.002189142
feature vectors 0.002185064
training time 0.002183111
likelihood training 0.002159035
feature changes 0.0021579150000000003
current training 0.002154033
features 0.00214957
descent training 0.0021359120000000002
single training 0.002121439
training process 0.0021031920000000003
training algo 0.00207192
sgd training 0.002060584
batch training 0.002036133
training times 0.0020308
fast training 0.002026561
whole training 0.002025788
efficient training 0.002019655
training samples 0.002014319
feature 0.00191466
current word 0.0019146130000000001
crf model 0.0018782170000000002
tagging model 0.001819876
data set 0.001777907
training 0.00176252
learning models 0.001660543
same data 0.0016566760000000002
test data 0.001637103
trained model 0.001614812
learning method 0.001589676
compact model 0.001589631
other words 0.0015258630000000001
ing data 0.001509618
linear models 0.001416068
recognition data 0.00138664
learning rate 0.0013547610000000001
crf models 0.001323435
weight vector 0.0013234229999999998
model 0.00131864
heldout data 0.0013036179999999999
ging data 0.001301551
test set 0.00129663
learning framework 0.001287546
machine learning 0.001260272
first set 0.001236724
learning algo 0.001206085
sgd learning 0.0011947490000000002
second set 0.00116372
actual learning 0.0011501150000000002
adaptive learning 0.001146619
learning rates 0.001146475
efficient algorithm 0.001145694
bfgs algorithm 0.0011417200000000001
attractive learning 0.001140422
probabilistic models 0.001139136
folos algorithm 0.001131764
first problem 0.001113536
simple method 0.0010957620000000001
tive function 0.001084521
neighboring words 0.001080974
current weight 0.0010763679999999999
descent method 0.001066383
third set 0.001057381
objective function 0.00104548
second problem 0.001040532
compact models 0.001034849
partition function 0.001022754
accurate models 0.001013878
text chunks 9.86492E-4
pos tagging 9.7968E-4
same level 9.744770000000001E-4
weight updates 9.7331E-4
same regularization 9.676770000000001E-4
other param 9.659320000000001E-4
other components 9.62065E-4
text chunk 9.59051E-4
tagging task 9.583739999999999E-4
naive method 9.56209E-4
true weight 9.48995E-4
ent method 9.439E-4
final accuracy 9.4362E-4
newton method 9.36115E-4
text classification 9.343530000000001E-4
vector machines 9.23694E-4
different strategy 9.23482E-4
support vector 9.177769999999999E-4
sample log 9.173709999999999E-4
text processing 9.155770000000001E-4
regularization parameter 9.131320000000001E-4
memory parameter 9.1184E-4
text chunking 9.09029E-4
ing rate 9.085040000000001E-4
learning 8.96685E-4
