training data 0.003391741
test data 0.003224813
joint model 0.003198911
first data 0.003023408
data set 0.002941076
data sets 0.002914279
state model 0.0028769299999999998
model size 0.002825602
tense data 0.0027634120000000002
second data 0.002760044
the data 0.002758541
german data 0.002757981
slovene data 0.0027552790000000002
conditional model 0.002751655
arabic data 0.0027412450000000002
ling data 0.0027384320000000003
nakisa data 0.0027330740000000003
ept data 0.0027258070000000002
mccarthy data 0.002712361
data point 0.0027049630000000003
data sparsity 0.0027041820000000003
generative model 0.00267256
model outputs 0.0026369089999999998
particular model 0.002614638
tional model 0.002596759
parametric model 0.002562344
model 0.00231161
learning approach 0.0017001960000000002
training algorithm 0.0016369660000000001
supervised learning 0.0016196420000000001
unsupervised learning 0.001595604
learning techniques 0.001526937
machine learning 0.001501713
generative models 0.00142803
training set 0.001421657
framework learning 0.001414034
symbolic learning 0.00139358
models states 0.0013859179999999999
joint probability 0.0013723799999999999
markov models 0.001362221
possible features 0.00134712
joint probabilities 0.00132509
feature vector 0.001284654
irregular words 0.0012815740000000002
new word 0.001241438
feature space 0.001232852
inflected words 0.001209743
conditional features 0.001205265
programming training 0.001184724
frequent words 0.001168192
nonce words 0.00115685
uninflected words 0.00115303
ing algorithm 0.001145009
scaled joint 0.0011448209999999999
learning 0.00114342
name information 0.0011426370000000002
joint probabil 0.001136282
different ways 0.001129936
estimation algorithm 0.001115367
label language 0.001113887
language source 0.001107066
different metrics 0.001094013
distance function 0.001088533
simple approach 0.00108386
large number 0.001082366
models 0.00106708
kernel function 0.001056254
other types 0.001052185
test ling 0.001052085
other cases 0.001048492
ditional features 0.001036624
same effect 0.00103427
other aspects 0.001032341
train test 0.001022321
phonological features 0.0010159980000000002
normal hidden 0.0010101490000000001
many languages 0.001003501
large part 0.001001772
input information 9.901E-4
current state 9.75721E-4
fisher information 9.72915E-4
irregular form 9.62993E-4
large majority 9.603369999999999E-4
plural form 9.58473E-4
scent algorithm 9.51735E-4
high performance 9.50067E-4
standard techniques 9.46493E-4
same symbol 9.44222E-4
training 9.36161E-4
regular form 9.35503E-4
method work 9.302520000000001E-4
approach scores 9.302500000000001E-4
state sequence 9.289210000000001E-4
conditional probability 9.25124E-4
normal dynamic 9.15407E-4
initial state 9.09734E-4
standard hmm 9.0933E-4
information geometry 9.07951E-4
end state 9.032770000000001E-4
words 9.01836E-4
