Noun
conditional entropy (plural conditional entropies)
(information theory) The portion of a random variable's own Shannon entropy which is independent from another, given, random variable.
The conditional entropy of random variable
Y
{\displaystyle Y}
given
X
{\displaystyle X}
(i.e., conditioned by
X
{\displaystyle X}
), denoted as
H
(
Y
|
X
)
{\displaystyle H(Y|X)}
, is equal to
H
(
Y
)
−
I
(
Y
;
X
)
{\displaystyle H(Y)-I(Y;X)}
where
I
(
Y
;
X
)
{\displaystyle I(Y;X)}
is the mutual information between
Y
{\displaystyle Y}
and
X
{\displaystyle X}
.