Information Technology Reference
In-Depth Information
Fig. 3.23 shows the evolution of the decision border and the error PDF
(computed only at the
e
i
values) at successive epochs. Fig. 3.24 shows the
converged solution. In the final epochs the radius of the circle undergoes the
main adjustments converging to a value close to
a
opt
.
Finally, Fig. 3.25 shows the learning curves for this experimental setting,
evidencing a consistent learning of the MEE-trained circle.
0.4
2
x
2
^
E
(e)
1
0.3
0
0.2
−1
x
1
−2
e
0.1
−2
0
2
−2
−1
0
1
2
(a) Epochs=0
0.4
x
2
2
^
E
(e)
1
0.3
0
0.2
−1
x
1
−2
e
0.1
−2
0
2
−2
−1
0
1
2
(b) Epochs=45
0.4
x
2
2
^
E
(e)
0.3
1
0
0.2
−1
0.1
x
1
−2
e
0
−2
−1
0
1
2
−2
0
2
(c) Epochs=48
Fig. 3.23 The ring dataset of Example 3.11 at different epochs of the MEE-trained
circle. The left graphs show the dataset with the circular decision border (solid line).
The right graphs show the error PDF in the
E
=[
−
2
,
2]
support.