Information Technology Reference
In-Depth Information
Table 11.1.
Summary of the dierent experiments and results.
Exp.
Parameters
E
1
A
ij
>= 0 (see Ref. 3)
9.69%
A
=
M
=
A
+
M
= 29:44
S
= 1
2
A
ij
>= 0
8.75%
A
=
A
= 26:10
M
=
M
= 3:34
S
= 1
3
A
ij
>= 0
10.15%
A
=
A
= 26:10
M
=
M
= 3:34
S
=
S
= 0:15
4
A
ij
2f0; 1g
8.33%
A
=
A
= 8:20
M
=
M
= 3:34
S
= 1
5
A
ij
2f0; 1g
9.62%
A
=
A
= 8:20
M
=
M
= 3:34
S
=
S
= 0:15
6
A
ij
replaced by relatedness R
ij
(lineage distance)
11.76%
A
=
A
= 2255:97
M
=
M
= 3:34
S
=
S
= 0:15
7
A
ij
= 1 ;8i; j
11.60%
A
=
A
= 139:60
M
=
M
= 3:34
S
=
S
= 0:15
11.4.8. Summary
The summary of the experiments results are presented in Table 11.1. These show
that the outcome of the optimization problems are basically driven by the informa-
tion we provide a priori, i.e. the position of the sensor and muscle neurons, which
determine the main anchorage of neurons' connections, with very low impact of
the neuron-neuron connectivity in the process. We also present a persuasive and
forceful comparison of the resulting layouts of the experiments at Fig. 11.10. Note
that the kernel of the prediction success is in the ventral cord neuron group (G7)
in the central area of the worm, which is not surprising given the biological nature
of these neurons, see [10]. The ventral cord is spatially coherent; neurons, partic-
ularly interneurons running in the cord, maintain their positions relative to their
nearest neighbors in spite of local distortions produced by intrusions of cell bodies,
