<s>
The	O
auxiliary	B-Algorithm
particle	I-Algorithm
filter	I-Algorithm
is	O
a	O
particle	B-Algorithm
filtering	I-Algorithm
algorithm	O
introduced	O
by	O
Pitt	O
and	O
Shephard	O
in	O
1999	O
to	O
improve	O
some	O
deficiencies	O
of	O
the	O
sequential	O
importance	O
resampling	O
(	O
SIR	O
)	O
algorithm	O
when	O
dealing	O
with	O
tailed	O
observation	O
densities	O
.	O
</s>
<s>
Particle	B-Algorithm
filters	I-Algorithm
approximate	O
continuous	O
random	O
variable	O
by	O
particles	O
with	O
discrete	O
probability	O
mass	O
,	O
say	O
for	O
uniform	O
distribution	O
.	O
</s>
<s>
The	O
particle	B-Algorithm
filters	I-Algorithm
draw	O
samples	O
from	O
the	O
prior	O
density	O
.	O
</s>
<s>
The	O
weakness	O
of	O
the	O
particle	B-Algorithm
filters	I-Algorithm
includes	O
:	O
</s>
<s>
Therefore	O
,	O
the	O
auxiliary	B-Algorithm
particle	I-Algorithm
filter	I-Algorithm
is	O
proposed	O
to	O
solve	O
this	O
problem	O
.	O
</s>
<s>
The	O
particle	B-Algorithm
filters	I-Algorithm
draw	O
samples	O
from	O
.	O
</s>
<s>
The	O
original	O
particle	B-Algorithm
filters	I-Algorithm
draw	O
samples	O
from	O
the	O
prior	O
density	O
,	O
while	O
the	O
auxiliary	O
filters	O
draw	O
from	O
the	O
joint	O
distribution	O
of	O
the	O
prior	O
density	O
and	O
the	O
likelihood	O
.	O
</s>
<s>
In	O
other	O
words	O
,	O
the	O
auxiliary	B-Algorithm
particle	I-Algorithm
filters	I-Algorithm
avoid	O
the	O
circumstance	O
which	O
the	O
particles	O
are	O
generated	O
in	O
the	O
regions	O
with	O
low	O
likelihood	O
.	O
</s>
