D14-1051 |
method has to be evaluated with the
|
Kullback-Liebler divergence
|
metric for each topic space .
|
W13-2232 |
is a symmetric version of the
|
Kullback-Liebler divergence
|
. For the JS approximation ,
|
N09-1053 |
is to find the nearest model in
|
Kullback-Liebler divergence
|
that satisfies a set of linear
|
W03-1201 |
till the sum of the squares of
|
Kullback-Liebler divergences
|
between CPTs in successive iterations
|
W12-3121 |
combination methods . We also report the
|
Kullback-Liebler divergence
|
( KL ) between the BLEU Oracle
|
W07-2216 |
are always guaranteed that the
|
Kullback-Liebler divergence
|
between two approximated distributions
|
W10-4106 |
language understanding , mainly using
|
Kullback-Liebler divergence
|
and mutual information . Pargellis
|
W12-0901 |
equation ( 5 ) is the well-known
|
Kullback-Liebler divergence
|
DKL ( M2 | | M1 ) of the two
|
W98-1122 |
acquisition is performed through
|
Kullback-Liebler divergence
|
techniques with application to
|
P13-1144 |
◦ grid cells and assigns
|
Kullback-Liebler divergences
|
to each cell given a document
|
D13-1116 |
original one , by minimizing the
|
Kullback-Liebler divergence
|
between the two -- see for instance
|
W00-0707 |
the values which minimize the
|
Kullback-Liebler divergence
|
D ( pliq ) between the model
|
P06-1035 |
use two measures : the symmetric
|
Kullback-Liebler divergence
|
( Jeffreys , 1946 ) and the Rao
|