<s>
A	O
learning	B-Algorithm
augmented	I-Algorithm
algorithm	I-Algorithm
is	O
an	O
algorithm	O
that	O
can	O
make	O
use	O
of	O
a	O
prediction	O
to	O
improve	O
its	O
performance	O
.	O
</s>
<s>
Whereas	O
in	O
regular	O
algorithms	O
just	O
the	O
problem	O
instance	O
is	O
inputted	O
,	O
learning	B-Algorithm
augmented	I-Algorithm
algorithms	I-Algorithm
accept	O
an	O
extra	O
parameter	O
.	O
</s>
<s>
A	O
learning	B-Algorithm
augmented	I-Algorithm
algorithm	I-Algorithm
typically	O
takes	O
an	O
input	O
.	O
</s>
<s>
Learning	B-Algorithm
augmented	I-Algorithm
algorithms	I-Algorithm
usually	O
satisfy	O
the	O
following	O
two	O
properties	O
:	O
</s>
<s>
A	O
learning	B-Algorithm
augmented	I-Algorithm
algorithm	I-Algorithm
is	O
said	O
to	O
be	O
consistent	O
if	O
the	O
algorithm	O
can	O
be	O
proven	O
to	O
have	O
a	O
good	O
performance	O
when	O
it	O
is	O
provided	O
with	O
an	O
accurate	O
prediction	O
.	O
</s>
<s>
Learning	B-Algorithm
augmented	I-Algorithm
algorithms	I-Algorithm
generally	O
do	O
not	O
prescribe	O
how	O
the	O
prediction	O
should	O
be	O
done	O
.	O
</s>
<s>
With	O
a	O
prediction	O
for	O
the	O
position	O
of	O
,	O
the	O
following	O
learning	B-Algorithm
augmented	I-Algorithm
algorithm	I-Algorithm
can	O
be	O
used	O
.	O
</s>
<s>
In	O
the	O
learning	B-Algorithm
augmented	I-Algorithm
algorithm	I-Algorithm
,	O
probing	O
the	O
positions	O
takes	O
steps	O
.	O
</s>
<s>
Learning	B-Algorithm
augmented	I-Algorithm
algorithms	I-Algorithm
are	O
known	O
for	O
:	O
</s>
