word corpus 0.003844149
dutch word 0.003301078
word representations 0.003150107
orthographic word 0.003086736
learning error 0.001761983
training corpus 0.001747177
random corpus 0.0015415110000000002
test corpus 0.00145839
local error 0.001451559
evaluation methods 0.001394562
input data 0.001379352
error function 0.001379074
previous methods 0.001336883
training input 0.0013353290000000001
corpus representation 0.001327499
prediction error 0.001318577
error functions 0.001311132
network evaluation 0.0012777499999999998
error rate 0.001263889
tolerable error 0.001236583
entire corpus 0.001231513
monosyllabic corpus 0.0012152040000000001
corpus representat 0.001213841
evaluation method 0.00120061
other approach 0.001180469
neural network 0.0011479419999999999
data structure 0.001145536
examination methods 0.001119486
dutch language 0.001095405
threshold method 0.0010926339999999999
valuation methods 0.001090954
sequential data 0.001087608
evaluations methods 0.001081959
diverse methods 0.001081959
output layer 0.001076993
language analysis 0.0010642310000000001
learning task 0.001054513
srn training 0.001042057
other languages 0.0010412730000000001
srn learning 0.001026502
network response 0.0010211929999999999
error 0.00101274
learning examination 0.0010121330000000001
distribution vector 0.0010053269999999999
connectionist language 0.001004369
recurrent network 9.93805E-4
input layer 9.84469E-4
corpus 9.82379E-4
other tasks 9.747530000000001E-4
trained language 9.70269E-4
natural language 9.652899999999999E-4
network responses 9.633479999999999E-4
context layer 9.59093E-4
different neuron 9.51679E-4
output neurons 9.47389E-4
probability distribution 9.4631E-4
srn output 9.40314E-4
same time 9.36024E-4
characters distribution 9.33102E-4
effective method 9.27928E-4
output neuron 9.26695E-4
possible approaches 9.16549E-4
different operations 9.14273E-4
additional information 9.094560000000001E-4
output activations 9.00812E-4
vector size 8.959549999999999E-4
words 8.93996E-4
language phonotaetles 8.87369E-4
language span 8.85234E-4
language proce 8.835620000000001E-4
possible successors 8.81943E-4
random corpora 8.776340000000001E-4
complete set 8.74213E-4
possible children 8.58352E-4
methods 8.56596E-4
response accuracy 8.39586E-4
empirical distribution 8.36981E-4
second evaluation 8.32581E-4
left context 8.325089999999999E-4
input sequences 8.25804E-4
srn evaluation 8.152249999999999E-4
second approach 8.130640000000001E-4
alternative evaluation 8.08392E-4
test procedure 8.045159999999999E-4
evaluation procedures 7.99569E-4
several successors 7.98908E-4
random strings 7.9763E-4
familiar input 7.942170000000001E-4
first case 7.87525E-4
context strings 7.836529999999999E-4
time complexity 7.80455E-4
empirical probability 7.75413E-4
context provirtea 7.69397E-4
final evaluation 7.67799E-4
training 7.64798E-4
instant evaluation 7.63412E-4
empirical characters 7.62205E-4
empirical probabilities 7.59879E-4
learning 7.49243E-4
simple recurrent 7.48048E-4
