Digital Signal Processing Reference
In-Depth Information
1.6.2 Bayes' Rule
Let the space
S
be subdivided into
N
mutually exclusive events
B
j
,
j ¼
1,
,
N
.
The probabilities of events
B
j
are called
a priori probabilities
because they repre-
sent the probabilities of
B
j
before the experiment is performed. Similarly, the
probabilities
P
{
A
|
B
j
} are typically known prior to conducting the experiment, and
are called
transition probabilities
.
Now, suppose that the experiment is performed and, as a result, event
A
occurred. The probability of the occurrence of any of the events
B
j
—knowing
that the event
A
has occurred-is called
a posteriori probability PfB
j
jAg
.Thea
posteriori probability is calculated using conditional probability (
1.43
) and the
theorem of total probability (
1.65
) resulting in:
...
P
f
AB
j
g
PfAg
¼
PfAjB
j
gPfB
j
g
P
PfB
j
jAg¼
:
(1.70a)
N
i¼
1
PfAjB
i
gPfB
i
g
The formula in (
1.70a
) is known as
Bayes' rule
.
Because (
1.70a
) is a complicated formula, some authors like Haddad [HAD06,
p. 32] consider it is better to write Bayes' rule using the following two equations:
P
f
AB
j
g
PfAg
¼
PfAjB
j
gPfB
j
g
PfAg
PfB
j
jAg¼
PfAg¼
X
N
PfAjB
i
gPfB
i
g:
(1.70b)
i¼
1
1.7
Independent Events
Consider two events
A
and
B
with the nonzero probabilities
P
{
A
} and
P
{
B
}, and
that the occurrence of one event does not affect the occurrence of the other event.
That means that the conditional probability of event
A
, given
B
, is equal to the
probability of event
A
,
PfAjBg¼PfAg:
(1.71)
Similarly, the conditional probability of event
B
, given
A
, is equal to the
probability of event
B
,
PfBjAg¼PfBg:
(1.72)
Search WWH ::
Custom Search