Graphics Reference
In-Depth Information
Figure
.
.
PCA and KPCA applied to a synthetic “two moons” data set: (a)originaldata,(b
)PCA
results, (b
) results from KPCA with the polynomial kernel of degree
, (b
) results from KPCA with
the Gaussian kernel, σ
. In each of these histograms (b
-b
), the horizontal axis is the principal
coordinate from either the PCA or KPCA and the vertical axis is the frequency. Plot (c)showstheROC
curves for these three procedures, with area under curve reported as
=
A
=
.
,
A
=
.
PCA
KPCA
(
Poly
)
and
A
=
.
KPCA
(
RBF
)
