<s>
They	O
include	O
both	O
model-based	B-General_Concept
methods	I-General_Concept
,	O
where	O
a	O
generative	O
model	O
is	O
available	O
or	O
can	O
be	O
learned	O
,	O
in	O
addition	O
to	O
model-free	B-General_Concept
methods	I-General_Concept
,	O
that	O
include	O
regression-based	O
approaches	O
,	O
such	O
as	O
stacked-regression	O
.	O
</s>
<s>
Bayesian	B-General_Concept
linear	I-General_Concept
regression	I-General_Concept
,	O
Gaussian	O
mixture	O
models	O
,	O
Gaussian	B-General_Concept
processes	I-General_Concept
,	O
auto-regressive	B-Algorithm
Gaussian	B-General_Concept
processes	I-General_Concept
,	O
or	O
Bayesian	O
polynomial	O
chaos	O
expansions	O
.	O
</s>
<s>
For	O
example	O
,	O
LoFi	O
data	O
can	O
be	O
produced	O
by	O
models	O
of	O
a	O
physical	O
system	O
that	O
use	O
approximations	B-Algorithm
to	O
simulate	O
the	O
system	O
,	O
rather	O
than	O
modeling	O
the	O
system	O
in	O
an	O
exhaustive	O
manner	O
.	O
</s>
<s>
For	O
example	O
,	O
low-fidelity	O
data	O
can	O
be	O
acquired	O
by	O
using	O
a	O
distributed	B-Operating_System
simulation	O
platform	O
,	O
such	O
as	O
X-Plane	O
,	O
and	O
requiring	O
novice	O
participants	O
to	O
operate	O
in	O
scenarios	O
that	O
are	O
approximations	B-Algorithm
of	O
the	O
real-world	O
context	O
.	O
</s>
<s>
In	O
an	O
auto-regressive	B-Algorithm
model	O
of	O
Gaussian	B-General_Concept
processes	I-General_Concept
(	O
GP	O
)	O
,	O
each	O
level	O
of	O
output	O
fidelity	O
,	O
,	O
where	O
a	O
higher	O
denotes	O
a	O
higher	O
fidelity	O
,	O
is	O
modeled	O
as	O
a	O
GP	O
,	O
,	O
which	O
can	O
be	O
expressed	O
in	O
terms	O
of	O
the	O
previous	O
level	O
's	O
GP	O
,	O
,	O
a	O
proportionality	O
constant	O
and	O
a	O
"	O
difference-GP	O
"	O
as	O
follows	O
:	O
</s>
