<s>
Domain	B-General_Concept
adaptation	I-General_Concept
is	O
a	O
field	O
associated	O
with	O
machine	O
learning	O
and	O
transfer	B-General_Concept
learning	I-General_Concept
.	O
</s>
<s>
Domain	B-General_Concept
adaptation	I-General_Concept
has	O
also	O
been	O
shown	O
to	O
be	O
beneficial	O
for	O
learning	O
unrelated	O
sources	O
.	O
</s>
<s>
Note	O
that	O
,	O
when	O
more	O
than	O
one	O
source	O
distribution	O
is	O
available	O
the	O
problem	O
is	O
referred	O
to	O
as	O
multi-source	O
domain	B-General_Concept
adaptation	I-General_Concept
.	O
</s>
<s>
Domain	B-General_Concept
adaptation	I-General_Concept
is	O
the	O
ability	O
to	O
apply	O
an	O
algorithm	O
trained	O
in	O
one	O
or	O
more	O
"	O
source	O
domains	O
"	O
to	O
a	O
different	O
(	O
but	O
related	O
)	O
"	O
target	O
domain	O
"	O
.	O
</s>
<s>
Domain	B-General_Concept
adaptation	I-General_Concept
is	O
a	O
subcategory	O
of	O
transfer	B-General_Concept
learning	I-General_Concept
.	O
</s>
<s>
In	O
domain	B-General_Concept
adaptation	I-General_Concept
,	O
the	O
source	O
and	O
target	O
domains	O
all	O
have	O
the	O
same	O
feature	O
space	O
(	O
but	O
different	O
distributions	O
)	O
;	O
in	O
contrast	O
,	O
transfer	B-General_Concept
learning	I-General_Concept
includes	O
cases	O
where	O
the	O
target	O
domain	O
's	O
feature	O
space	O
is	O
different	O
from	O
the	O
source	O
feature	O
space	O
or	O
spaces	O
.	O
</s>
<s>
The	O
modern	O
machine-learning	O
community	O
has	O
many	O
different	O
strategies	O
to	O
attempt	O
to	O
gain	O
better	O
domain	B-General_Concept
adaptation	I-General_Concept
.	O
</s>
<s>
Other	O
applications	O
include	O
wifi	O
localization	O
detection	O
and	O
many	O
aspects	O
of	O
computer	B-Application
vision	I-Application
.	O
</s>
<s>
Usually	O
in	O
supervised	B-General_Concept
learning	I-General_Concept
(	O
without	O
domain	B-General_Concept
adaptation	I-General_Concept
)	O
,	O
we	O
suppose	O
that	O
the	O
examples	O
are	O
drawn	O
i.i.d.	O
</s>
<s>
The	O
main	O
difference	O
between	O
supervised	B-General_Concept
learning	I-General_Concept
and	O
domain	B-General_Concept
adaptation	I-General_Concept
is	O
that	O
in	O
the	O
latter	O
situation	O
we	O
study	O
two	O
different	O
(	O
but	O
related	O
)	O
distributions	O
and	O
on	O
.	O
</s>
<s>
The	O
domain	B-General_Concept
adaptation	I-General_Concept
task	O
then	O
consists	O
of	O
the	O
transfer	O
of	O
knowledge	O
from	O
the	O
source	O
domain	O
to	O
the	O
target	O
one	O
.	O
</s>
<s>
There	O
are	O
several	O
contexts	O
of	O
domain	B-General_Concept
adaptation	I-General_Concept
.	O
</s>
<s>
The	O
unsupervised	O
domain	B-General_Concept
adaptation	I-General_Concept
:	O
the	O
learning	O
sample	O
contains	O
a	O
set	O
of	O
labeled	O
source	O
examples	O
,	O
a	O
set	O
of	O
unlabeled	O
source	O
examples	O
and	O
a	O
set	O
of	O
unlabeled	O
target	O
examples	O
.	O
</s>
<s>
The	O
semi-supervised	O
domain	B-General_Concept
adaptation	I-General_Concept
:	O
in	O
this	O
situation	O
,	O
we	O
also	O
consider	O
a	O
"	O
small	O
"	O
set	O
of	O
labeled	O
target	O
examples	O
.	O
</s>
<s>
The	O
supervised	O
domain	B-General_Concept
adaptation	I-General_Concept
:	O
all	O
the	O
examples	O
considered	O
are	O
supposed	O
to	O
be	O
labeled	O
.	O
</s>
<s>
This	O
can	O
be	O
achieved	O
through	O
the	O
use	O
of	O
Adversarial	B-General_Concept
machine	I-General_Concept
learning	I-General_Concept
techniques	O
where	O
feature	O
representations	O
from	O
samples	O
in	O
different	O
domains	O
are	O
encouraged	O
to	O
be	O
indistinguishable	O
.	O
</s>
<s>
Several	O
compilations	O
of	O
domain	B-General_Concept
adaptation	I-General_Concept
and	O
transfer	B-General_Concept
learning	I-General_Concept
algorithms	O
have	O
been	O
implemented	O
over	O
the	O
past	O
decades	O
:	O
</s>
