feature learning 0.003514296
feature set 0.003085458
structured feature 0.003005864
standard feature 0.002923611
form feature 0.002903058
feature space 0.002900156
several feature 0.002894921
single feature 0.002893546
feature template 0.002849121
feature dropout 0.002846106
original feature 0.002837095
noise feature 0.002834065
efficient feature 0.002823461
feature structure 0.002818712
individual feature 0.0028184630000000002
feature templates 0.002776322
transition feature 0.002759999
feature spaces 0.002752694
feature projection 0.002745526
feature projections 0.002739107
feature transformation 0.002738923
havek feature 0.00273801
words features 0.002632224
feature 0.00252652
other features 0.002507567
structured features 0.002421544
new features 0.002330734
active features 0.002263681
original features 0.0022527750000000003
noise features 0.002249745
base features 0.002225243
binary features 0.002186801
pivot features 0.002160941
rare features 0.0021543590000000002
ternative features 0.0021543590000000002
features 0.0019422
learning algorithm 0.001624941
training data 0.001493628
representation learning 0.00145176
learning methods 0.001385184
deep learning 0.0013588970000000002
correspondence learning 0.0012218370000000001
training set 0.001220642
learning community 0.0012016460000000002
spondence learning 0.00120103
different domain 0.0011754880000000001
labeled data 0.001143224
work domain 0.001085123
development data 0.001084694
dimensional data 0.001080729
ment data 0.001078533
corrupted data 0.001077386
model 0.00107283
previous word 0.001068717
learning 9.87776E-4
current word 9.86918E-4
weight matrix 9.64409E-4
different values 9.621569999999999E-4
domain adaptation 9.53501E-4
target domain 9.33871E-4
structured fea 9.23381E-4
unsupervised domain 9.01565E-4
classification tasks 8.945909999999999E-4
source domain 8.9435E-4
full set 8.71838E-4
previous work 8.707680000000001E-4
other domains 8.63251E-4
other entries 8.62782E-4
different types 8.572930000000001E-4
similar results 8.515E-4
same template 8.502749999999999E-4
domain adap 8.5014E-4
several representation 8.32385E-4
such autoencoders 8.290719999999999E-4
such approaches 8.21117E-4
test sets 8.18582E-4
structured dropout 7.9893E-4
different approaches 7.96019E-4
corpus crf 7.9113E-4
different templates 7.888999999999999E-4
alternative value 7.6994E-4
nlp tasks 7.6661E-4
standard denoising 7.61374E-4
brahe corpus 7.57441E-4
noise distributions 7.56954E-4
newstext corpus 7.52284E-4
tion tasks 7.494089999999999E-4
results table 7.4867E-4
sequence labeling 7.307710000000001E-4
tion matrix 7.28761E-4
standard dropout 7.16677E-4
adaptation methods 7.14519E-4
many nlp 7.100979999999999E-4
uniform distribution 7.09197E-4
tent representation 7.07465E-4
structured scambling 6.95029E-4
new technique 6.92274E-4
words 6.90024E-4
novel types 6.869490000000001E-4
tactic tasks 6.82936E-4
