Biomedical Engineering Reference
In-Depth Information
a
b
1
0.5
0
1
0.5
0
-0.5
-1
1
0.5
0
-0.5
-1
1
0.5
0
-0.5
-1
100
200
300
400
500
600
700
800
900
1000
-0.5
-1
1
0.5
0
-0.5
-1
1
0.5
0
-0.5
-1
0 100 200
300
400 500
0 100 200
300
400 500
0100 200
300
400 500
1
0.5
0
-0.5
-1
0 100 200
300
400 500
0 100 200
300
400 500
0100 200
300
400 500
1
0.5
0
-0.5
-1
1
0.5
0
-0.5
-1
0 100 200
300
400 500
0 100 200
300
400 500
0100 200
300
400 500
50 100 150 200 250 300 350 400 450
c
d
8
6
4
2
0
-2
-4
-6
1
0.5
0
-0.5
-1
-1.5
-2
8
6
4
8
6
2
4
0
-8
-6
-4
-2
0
2
4
6
8
2
-2
0
-4
-2
-4
-6
-6
-8
-8
Fig. 7.3
PCA analysis of a set of 1,000 jittered synthetic time series of 512 time samples. (
a
)Nine
time series out of 1,000. (
b
) Original raster plot. (
c
) 2D PCA projection. (
d
) 3D PCA projection
two points
x
and
z
are close according to the manifold
M
,then
f
(
x
)
and
f
(
z
)
must
n
. To express this regularity constraint in mathematical terms,
also be close in
R
notice that for
n
=1
, a Taylor expansion provides the following inequality [
2
]
|
f
(
z
)
−
f
(
x
)
|≤
d
M
(
x, z
)
||∇
f
(
x
)
||
+
o
(
d
M
(
x, z
))
,
where
∇
f
stands for the gradient of
f
and
d
M
is the geodesic distance on the
manifold between points
x
and
z
. The notation
g
(
z
)=
o
(
d
M
(
x, z
))
means that
)
d
M
(
x,z
)
g
(
z
as
z
tends towards
x
.
In order to obtain an embedding that satisfies the
“regularity”
constraint,
Laplacian-based methods control the smoothness of
f
globally by minimizing
0
tends to
M
||∇
f
(
x
)
||
2
p
(
x
)
dx
, provided that
M
||
)
||
2
p
f
(
x
(
x
)
dx
=1
.
(7.3)
The latter condition removes the scaling indeterminacy for the function
f
.
Search WWH ::
Custom Search