same feature 0.002337735
feature vector 0.002215806
feature noising 0.002059333
feature noise 0.002053455
dropout feature 0.002043021
single feature 0.00198787
feature sparsity 0.001987785
local feature 0.00197818
feature vectors 0.001971268
noised feature 0.0019271730000000001
active feature 0.0019221190000000001
cial feature 0.001920274
feature vec 0.0019069920000000001
feature correlations 0.0018995210000000002
feature expecta 0.0018995210000000002
simulated feature 0.0018995210000000002
crf model 0.001745564
word set 0.0016882540000000001
feature 0.00166306
ing features 0.001651137
test features 0.001633715
original features 0.001632682
model expectations 0.001630944
important features 0.001602757
node features 0.001578297
training data 0.0015710799999999999
input features 0.0015624060000000001
actual model 0.0015566970000000001
model prob 0.001554412
model uncertainty 0.001553339
sparse features 0.001536512
tive features 0.001525744
noised features 0.0015188530000000001
active features 0.0015137990000000001
rare features 0.001512576
training set 0.001509023
transition features 0.001504385
indicator features 0.001501429
corrupted features 0.001491764
sequence models 0.001461867
classification data 0.001423786
label sequence 0.001340333
previous word 0.001298855
model 0.00129816
tag sequence 0.001281636
set performance 0.001272136
features 0.00125474
labeled data 0.001228123
current word 0.0012274500000000002
test data 0.001211366
crf training 0.001186093
unlabeled data 0.001175174
set accuracy 0.001174813
word shape 0.0011658060000000001
word window 0.001165673
word bigram 0.001156884
word unigrams 0.0011565920000000001
disjunctive word 0.001155006
nlp models 0.001154949
test set 0.001149309
objective function 0.001145489
regularization tag 0.0011429650000000001
training example 0.0011424949999999999
training examples 0.001142348
real data 0.0011300820000000001
same crf 0.001122079
dropout training 0.00111865
original training 0.0011166309999999999
fake data 0.001109631
noised data 0.001096504
classification performance 0.001093197
text classification 0.001088864
training dataset 0.001075836
quadratic function 0.001072223
tropy models 0.001062283
gaussian method 0.001040275
development set 0.001032978
train set 0.001031094
tag pair 0.001028158
dev set 0.001018153
fake training 0.001015929
sequence tagging 0.001013789
velopment set 0.001006727
hensive set 0.001006727
label probabilities 9.889550000000001E-4
classification task 9.82882E-4
nlp sequence 9.81346E-4
tag precision 9.80317E-4
same order 9.78952E-4
crf inference 9.70478E-4
same datasets 9.6504E-4
conditional expectation 9.53353E-4
classification tasks 9.52863E-4
same trick 9.48445E-4
same distribu 9.350700000000001E-4
parameter esti 9.336710000000001E-4
same ner 9.31585E-4
sequence labeling 9.28157E-4
weight vector 9.201890000000001E-4
sequence length 9.126189999999999E-4
